pairs.yaml

abc_headlines:
- generation: the year they were published
  applications:
  - target: which categories of news dominated the headlines from year to year
    user: a historian writing about trends
    example_hypotheses:
    - talks about healthcare and medical breakthroughs or issues
    - mention sporting events, such as AFL games and the Rugby World Cup
    - focus on issues related to agriculture, such as the rise of female farmers and
      push for higher milk prices
    v2-origid: 3
    purely_exploratory: false
  pair_type: time
  pairs:
    '2007_2008':
      pos_desc: are ABC news headlines from 2007
      neg_desc: are ABC news headlines from 2008
      pos_class:
      - '2007'
      neg_class:
      - '2008'
      example_hypotheses:
      - talks less about the economy
      - talks less about the election
      - mentions President Bush
      - discusses terrorism
    '2010_2014':
      pos_desc: are ABC news headlines from 2010
      neg_desc: are ABC news headlines from 2014
      pos_class:
      - '2010'
      neg_class:
      - '2014'
      example_hypotheses:
      - mentions Obamacare
      - talks about the recession
    '2016_2017':
      pos_desc: are ABC news headlines from 2017
      neg_desc: are ABC news headlines from 2016
      pos_class:
      - '2017'
      neg_class:
      - '2016'
      example_hypotheses:
      - mentions Clinton
      - mentions polling
      - refers to Trump as a candidate
      - talks about Russia
    '2019_2020':
      pos_desc: are ABC news headlines from 2019
      neg_desc: are ABC news headlines from 2020
      pos_class:
      - '2019'
      neg_class:
      - '2020'
      example_hypotheses:
      - talks less about COVID-19
      - is more optimistic
      - mentions President Trump
      - does not mention Biden
  flip: true
ad_transcripts:
- generation: the industry of the company
  applications:
  - target: what different industries appeal to in ads
    user: an advertiser trying to learn about other industries
    example_hypotheses:
    - emphasizes the results of the product, such as long-lasting colour or healthier-looking
      skin
    - uses adjectives such as 'rich', 'sultry', and 'gorgeous' to describe the product
    - mentions natural ingredients and botanicals
    v2-origid: 5
    purely_exploratory: false
  pair_type: subject
  pairs:
    automative_travel:
      pos_desc: are ad transcripts for automobile companies
      neg_desc: are ad transcripts for travel companies
      pos_class:
      - Automotive
      neg_class:
      - Travel
      example_hypotheses:
      - talks more about cars
      - talks more about fuel efficiency
      - discusses speed
      - appeals to men
    beauty_care:
      pos_desc: are ad transcripts for beauty products
      neg_desc: are ad transcripts for personal care products
      pos_class:
      - Beauty
      neg_class:
      - Home & Personal Care
      example_hypotheses:
      - talks more about perception
      - talks more about fashion
      - appeals to women
      - mentions beauty brands
  flip: true
admin_statements:
- generation: the president leading the administration
  applications:
  - target: the legislative priorities of each administration
    user: a political scientist analyzing policy stances
    example_hypotheses:
    - mentions the Affordable Care Act
    - mentions the Dodd-Frank Wall Street Reform and Consumer Protection Act
    - acknowledges the importance of the African Growth and Opportunity Act
    v2-origid: 6
    purely_exploratory: false
  - target: the ideological stances of each administration
    user: a political scientist analyzing policy stances
    example_hypotheses:
    - discusses the importance of collaboration with States, Tribes, local governments,
      and private entities
    - highlights the need for investments in defense readiness and modernization
    - discusses the need for environmental justice
    v2-origid: 2006
    purely_exploratory: false
  pair_type: authorship
  pairs:
    obama_biden:
      pos_desc: are administration statements from Obama
      neg_desc: are administration statements from Biden
      pos_class:
      - 44-Obama
      neg_class:
      - 46-Biden
      example_hypotheses:
      - talks more about the Great Recession
      - does not mention COVID-19
      - mentions healthcare
      - discusses troops
      - mentions unemployment
    trump_biden:
      pos_desc: are administration statements from Trump
      neg_desc: are administration statements from Biden
      pos_class:
      - 45-Trump
      neg_class:
      - 46-Biden
      example_hypotheses:
      - talks more about immigration
      - favors conservative policies
      - talks less about COVID-19
      - is more hostile
  flip: true
airline_reviews:
- generation: the stars given in the review
  applications:
  - target: how to improve specific aspects of the airport and airplane experience
      for customers
    user: the manager of an airport
    example_hypotheses:
    - mentions the lack of a complimentary item (e.g. a muffin)
    - mentions the seat pitch being too small for a tall person
    - mentions the lack of USB and power outlets at the seat
    v2-origid: 12
    purely_exploratory: false
  pair_type: sentiment
  pairs:
    airline_comfort_rating_1_5:
      pos_desc: rate an airline's comfort 1/5
      neg_desc: rate an airline's comfort 5/5
      pos_class:
      - airline_overall_rating_1
      neg_class:
      - airline_overall_rating_5
      example_hypotheses:
      - talks about seat space
      - mentions uncomfortable seats
      - talks about the poor service
    airline_entertainment_rating_1_5:
      pos_desc: rate an airline's entertainment 1/5
      neg_desc: rate an airline's entertainment 5/5
      pos_class:
      - airline_entertainment_rating_1
      neg_class:
      - airline_entertainment_rating_5
      example_hypotheses:
      - complains about lack of wifi
      - mentions lack of selection of movies
      - talks about children being bored
    airline_food_rating_1_5:
      pos_desc: rate an airline's food 1/5
      neg_desc: rate an airline's food 5/5
      pos_class:
      - airline_food_rating_1
      neg_class:
      - airline_food_rating_5
      example_hypotheses:
      - talks about terrible tasting food
      - mentions food poisoning
    airline_overall_1_5:
      pos_desc: rate an airline overall 1/10
      neg_desc: rate an airline overall 5/10
      pos_class:
      - airline_overall_rating_1
      neg_class:
      - airline_overall_rating_5
      example_hypotheses:
      - is very frustrated
      - uses profanity
      - mentions a failure to receive a refund
      - mentions a canceled flight
    airline_overall_5_8:
      pos_desc: rate an airline overall 8/10
      neg_desc: rate an airline overall 10/10
      pos_class:
      - airline_overall_rating_8
      neg_class:
      - airline_overall_rating_10
      example_hypotheses:
      - mentions minor inconveniences
      - talks about discomfort during the flight
      - is neutral in tone
    airline_service_rating_1_5:
      pos_desc: rate an airline's service 1/5
      neg_desc: rate an airline's service 5/5
      pos_class:
      - airline_service_rating_1
      neg_class:
      - airline_service_rating_5
      example_hypotheses:
      - complains about untimely services
      - mentions lack of food
    airline_staff_rating_1_5:
      pos_desc: rate an airline's staff 1/5
      neg_desc: rate an airline's staff 5/5
      pos_class:
      - airline_staff_rating_1
      neg_class:
      - airline_staff_rating_5
      example_hypotheses:
      - talks about rude staff
      - mentions understaffing
    airline_value_rating_1_5:
      pos_desc: rate an airline's value 1/5
      neg_desc: rate an airline's value 5/5
      pos_class:
      - airline_value_rating_1
      neg_class:
      - airline_value_rating_5
      example_hypotheses:
      - talks about expensive pricing
      - mentions unexpected charges
    airport_cleanliness_rating_1_5:
      pos_desc: rate an airport's cleanliness 1/5
      neg_desc: rate an airport's cleanliness 5/5
      pos_class:
      - airport_cleanliness_rating_1
      neg_class:
      - airport_cleanliness_rating_5
      example_hypotheses:
      - mentions dirt on the floor
      - mentions food left on the seat
    airport_overall_1_5:
      pos_desc: rate an airport 1/10 overall
      neg_desc: rate an airport 5/10 overall
      pos_class:
      - airport_overall_rating_1
      neg_class:
      - airport_overall_rating_5
      example_hypotheses:
      - is very frustrated
      - mentions a terrible experience
    airport_overall_5_8:
      pos_desc: rate an airport 8/10 overall
      neg_desc: rate an airport 10/10 overall
      pos_class:
      - airport_overall_rating_8
      neg_class:
      - airport_overall_rating_10
      example_hypotheses:
      - mentions minor inconveniences
      - talks about the service or entertainment
    airport_queue_rating_1_5:
      pos_desc: rate an airport's queue 1/5
      neg_desc: rate an airport's queue 5/5
      pos_class:
      - airport_queue_rating_1
      neg_class:
      - airport_queue_rating_5
      example_hypotheses:
      - talks about long waits
      - complains about TSA
      - mentions missing a flight
    airport_shopping_rating_1_5:
      pos_desc: rate an airport's shopping avenues 1/5
      neg_desc: rate an airport's shopping avenues 5/5
      pos_class:
      - airport_shopping_rating_1
      neg_class:
      - airport_shopping_rating_5
      example_hypotheses:
      - talks about a lack of food
      - complains about pricing
      - mentions essential goods
  flip: false
- generation: the airline the reviewer flew
  applications:
  - target: which aspects of each airline stand out to customers
    user: a consumer researcher at an airline company
    example_hypotheses:
    - expresses frustration with the airline's refusal to offer any restitution for
      lost hotel reservations
    - mentions the staff as being unhelpful or rude
    - mentions difficulty with checking in or printing boarding passes
    v2-origid: 29
    purely_exploratory: false
  - target: what services and products are only offered by the other airline
    user: the CEO of an airline company
    example_hypotheses:
    - notes the Indian veg/non-veg meal on board
    - mentions wifi is not available in DXB
    - mentions the lack of a tablet for controlling the entertainment system
    v2-origid: 28
    purely_exploratory: false
  pair_type: subject
  pairs:
    american_delta:
      pos_desc: review American Airlines flights
      neg_desc: review Delta Airlines flights
      pos_class:
      - airline_american
      neg_class:
      - airline_delta
      example_hypotheses:
      - talks about higher prices
    british_ryan:
      pos_desc: review of British Airways flights
      neg_desc: review of Ryan Air flights
      pos_class:
      - airline_british
      neg_class:
      - airline_ryan
      example_hypotheses:
      - praises service
      - complains about cost
    canada:
      pos_desc: review Canada Airlines flights
      neg_desc: review Canada Airlines Rogue flights
      pos_class:
      - airline_canada
      neg_class:
      - airline_canada_rogue
      example_hypotheses:
      - does not mention business class
      - talks about worse service
    jet_emirates:
      pos_desc: review Jet Airways flights
      neg_desc: review Emirates flights
      pos_class:
      - airline_jet
      neg_class:
      - airline_emirates
      example_hypotheses:
      - mentions Dubai
      - talks about food options
    spirit_frontier:
      pos_desc: review Spirit flights
      neg_desc: review Frontier flights
      pos_class:
      - airline_spirit
      neg_class:
      - airline_frontier
      example_hypotheses:
      - complains about delayed flights
      - mentions pricing
    united_american:
      pos_desc: review United Airlines flights
      neg_desc: review American Airlines flights
      pos_class:
      - airline_united
      neg_class:
      - airline_american
      example_hypotheses:
      - talks about destinations outside of the US
      - mentions long flights
      - talks about food
  flip: true
- generation: the country the reviewer is from
  applications:
  - target: which specific features reviewers from each country care about
    user: an owner of an international airline
    example_hypotheses:
    - mentions the quality of the onboard food
    - mentions the price of the ticket
    - mentions the IFE and its various offerings
    v2-origid: 36
    purely_exploratory: false
  pair_type: demographic
  pairs:
    author_sg_in:
      pos_desc: are airline reviews from Singapore
      neg_desc: are airline reviews from India
      pos_class:
      - author_sg
      neg_class:
      - author_in
      example_hypotheses:
      - talks about service
      - mentions communication issues
      - mentions Asian food
    author_uk_us:
      pos_desc: are airline reviews from the UK
      neg_desc: are airline reviews from the US
      pos_class:
      - author_uk
      neg_class:
      - author_us
      example_hypotheses:
      - uses a condescending tone
      - is rude
    author_aus_cad:
      pos_desc: are airline reviews from Australia
      neg_desc: are airline reviews from Canada
      pos_class:
      - author_aus
      neg_class:
      - author_cad
      example_hypotheses:
      - is more critical
      - uses less polite language
    author_ger_fr:
      pos_desc: are airline reviews from Germany
      neg_desc: are airline reviews from France
      pos_class:
      - author_ger
      neg_class:
      - author_fr
      example_hypotheses:
      - is more critical
  flip: true
- generation: the type of seat on the reviewed plane
  applications:
  - target: what customers like and don't like about each type of seat
    user: a designer of airport seats
    example_hypotheses:
    - mentions difficulty getting out of the seat if the passenger in front reclines
    - comments on the lack of PTV or AVOD
    - describes the lack of space for cabin bags underneath the seat
    v2-origid: 38
    purely_exploratory: false
  pair_type: subject
  pairs:
    seats_3x3x3_3x3:
      pos_desc: review airplane seats in a 3x3x3 arrangement
      neg_desc: review airplane seats in a 3x3 arrangement
      pos_class:
      - seats_3x3x3
      neg_class:
      - seats_3x3
      example_hypotheses:
      - complains about room
      - mentions difficulty reaching bathrooms
    seats_3x4x3_2x4x2:
      pos_desc: review airplane seats in a 3x4x3 arrangement
      neg_desc: review airplane seats in a 2x4x2 arrangement
      pos_class:
      - seats_3x4x3
      neg_class:
      - seats_2x4x2
      example_hypotheses:
      - complains about being in the middle seat
      - talks about poor seatmates
  flip: true
- generation: the model of the plane the reviewer flew on
  applications:
  - target: what customers like and don't like about each plane model
    user: a product manager at an airplane manufacturer
    example_hypotheses:
    - mentions how the reclined seat in front makes it difficult to access the tray
      table
    - mentions how the entertainment screen is too close to the face to be visible
    - mentions how the entertainment system is excellent and includes USB charger
      ports
    v2-origid: 40
    purely_exploratory: false
  pair_type: subject
  pairs:
    seats_a340_a330:
      pos_desc: review seats on the Airbus 340
      neg_desc: review seats on the Airbus 330
      pos_class:
      - seats_a340
      neg_class:
      - seats_a330
      example_hypotheses:
      - mentions wider seats
      - more positive
      - praises the entertainment
    seats_a380_a340:
      pos_desc: review seats on the Airbus 380
      neg_desc: review seats on the Airbus 340
      pos_class:
      - seats_a380
      neg_class:
      - seats_a340
      example_hypotheses:
      - mentions wider seats
      - more positive
      - praises the entertainment
    seats_b777_b747:
      pos_desc: review seats on the Boeing 777
      neg_desc: review seats on the Boeing 747
      pos_class:
      - seats_b777
      neg_class:
      - seats_b747
      example_hypotheses:
      - mentions wider seats
      - more positive
      - praises the entertainment
  flip: true
- generation: the cabin of the customer
  applications:
  - target: the specific needs of customers in each cabin
    user: a product manager at an airline company
    example_hypotheses:
    - describes the difficulty of getting out of a row in-flight
    - mentions the need to haggle for dinner vouchers
    - mentions the lack of priority boarding and/or lounge access
    v2-origid: 42
    purely_exploratory: false
  pair_type: demographic
  pairs:
    seats_econ_prem:
      pos_desc: review airplane seats for economy passengers
      neg_desc: review airplane seats for premium passengers
      pos_class:
      - seats_econ
      neg_class:
      - seats_prem
      example_hypotheses:
      - complains about room
      - talks about poor service
  flip: true
- generation: the type of traveller
  applications:
  - target: the specific needs of different kinds of travellers
    user: a product manager at an airline company
    example_hypotheses:
    - mentions the high prices for food and drinks
    - complains about a lack of seating and a lack of wheelchair accommodations
    - mentions feeling rushed or not having enough time to make a connecting flight
    v2-origid: 43
    purely_exploratory: false
  pair_type: demographic
  pairs:
    traveller_family_business:
      pos_desc: are airline ratings from family travellers
      neg_desc: are airline ratings from business travellers
      pos_class:
      - traveller_family
      neg_class:
      - traveller_business
      example_hypotheses:
      - mentions food
      - mentions bathrooms
      - does not mention delays
    traveller_solo_couple:
      pos_desc: are airline ratings from solo travellers
      neg_desc: are airline ratings from couple travellers
      pos_class:
      - traveller_solo
      neg_class:
      - traveller_couple
      example_hypotheses:
      - focuses on practical needs
      - talks
  flip: true
aita:
- generation: whether Reddit commenters said they were in the wrong
  applications:
  - target: the speaking style of people in the wrong
    user: a sociologist studying moral judgements
    example_hypotheses:
    - employs guilt-tripping language, such as 'sacrifice' and 'heartbroken'
    - uses language to diminish the other person's opinion, such as 'completely unreasonable'
      or 'ridiculous'
    - uses generalizations or stereotypes about people
    v2-origid: 61
    purely_exploratory: false
  - target: the actions which were judged as right or wrong
    user: a sociologist studying moral judgements
    example_hypotheses:
    - demonstrates an effort to compromise or come to a mutual understanding
    - acknowledges their own mistakes and takes responsibility for their actions
    - demonstrates an understanding that there may be multiple perspectives on a given
      situation
    v2-origid: 50
    purely_exploratory: false
  pair_type: judgement
  pairs:
    verdict_a_nta:
      pos_desc: describe scenarios where the author is in the wrong
      neg_desc: describe scenarios where the author is not in the wrong
      pos_class:
      - verdict_a
      neg_class:
      - verdict_nta
      example_hypotheses:
      - is condescending in tone
      - doesn't express regret
      - is too confident
      - mentions making assumptions about other people
    verdict_es_a:
      pos_desc: describe scenarios where everyone is in the wrong
      neg_desc: describe scenarios where the author is in the wrong
      pos_class:
      - verdict_es
      neg_class:
      - verdict_a
      example_hypotheses:
      - describes multiple difficult people
      - talks about fighting
      - is condescending
    verdict_nah_nta:
      pos_desc: describe scenarios where no one is in the wrong
      neg_desc: describe scenarios where the other person is in the wrong
      pos_class:
      - verdict_nah
      neg_class:
      - verdict_nta
      example_hypotheses:
      - talks about a dilemma
      - expresses humility
    topic_work:
      pos_desc: describe work-related scenarios where the author is in the wrong
      neg_desc: describe work-related scenarios where the author is not in the wrong
      pos_class:
      - topic_work_is_asshole
      neg_class:
      - topic_work_not_asshole
      example_hypotheses:
      - describes failure to communicate
      - talks about a mean boss
      - mentions deadlines
    topic_sex:
      pos_desc: describe intercourse-related scenarios where the author is in the
        wrong
      neg_desc: describe intercourse-related scenarios where the author is not in
        the wrong
      pos_class:
      - topic_sex_is_asshole
      neg_class:
      - topic_sex_not_asshole
      example_hypotheses:
      - describes deception
      - mentions consent
      - is written by a man
    topic_ex:
      pos_desc: describe scenarios involving a former partner where the author is
        in the wrong
      neg_desc: describe scenarios involving a former partner where the author is
        in the wrong
      pos_class:
      - topic_ex_is_asshole
      neg_class:
      - topic_ex_not_asshole
      example_hypotheses:
      - describes resentment
      - talks about revenge
    topic_race:
      pos_desc: describe scenarios involving racism where the author is in the wrong
      neg_desc: describe scenarios involving racism where the author is not in the
        wrong
      pos_class:
      - topic_race_is_asshole
      neg_class:
      - topic_race_not_asshole
      example_hypotheses:
      - implies a shallow understanding of racism
      - talks about a joke
      - is written by someone who is white
    topic_husband:
      pos_desc: describe scenarios about the author's husband where the author is
        in the wrong
      neg_desc: describe scenarios about the author's husband where the author is
        not in the wrong
      pos_class:
      - topic_husband_is_asshole
      neg_class:
      - topic_husband_not_asshole
      example_hypotheses:
      - describes failure to communicate
      - talks about misreading emotions
    topic_wife:
      pos_desc: describe scenarios about the author's wife where the author is in
        the wrong
      neg_desc: describe scenarios about the author's wife where the author is not
        in the wrong
      pos_class:
      - topic_husband_is_asshole
      neg_class:
      - topic_husband_not_asshole
      example_hypotheses:
      - describes failure to communicate
      - mentions lack of support
    topic_sexuality:
      pos_desc: describe scenarios about sexuality where the author is in the wrong
      neg_desc: describe scenarios about sexuality where the author is not in the
        wrong
      pos_class:
      - topic_sexuality_is_asshole
      neg_class:
      - topic_sexuality_not_asshole
      example_hypotheses:
      - describes failure to communicate
      - suggests homophobia
    topic_children:
      pos_desc: describe scenarios about children where the author is in the wrong
      neg_desc: describe scenarios about children where the author is not in the wrong
      pos_class:
      - topic_children_is_asshole
      neg_class:
      - topic_children_not_asshole
      example_hypotheses:
      - implies hatred of children
      - mentions irresponsible actions
    topic_social_media:
      pos_desc: describe scenarios about social media where the author is in the wrong
      neg_desc: describe scenarios about social media where the author is not in the
        wrong
      pos_class:
      - topic_social_media_is_asshole
      neg_class:
      - topic_social_media_not_asshole
      example_hypotheses:
      - talks about a violation of trust
    topic_alcohol:
      pos_desc: describe scenarios involving alcohol where the author is in the wrong
      neg_desc: describe scenarios involving alcohol where the author is not in the
        wrong
      pos_class:
      - topic_alcohol_is_asshole
      neg_class:
      - topic_alcohol_not_asshole
      example_hypotheses:
      - describes failure to communicate
  flip: true
armenian_jobs:
- generation: the year the application was posted
  applications:
  - target: how the application requirements have evolved over time
    user: a journalist writing about the job market
    example_hypotheses:
    - asks for a cover letter
    - mentions the need for a letter of recommendation
    - requires citizenship
    v2-origid: 72
    purely_exploratory: false
  pair_type: time
  pairs:
    apps_pre_post_2012:
      pos_desc: are job applications requirements from 2010 to 2012
      neg_desc: are job applications requirements from 2013 to 2014
      pos_class:
      - app_process_years_2010_2013
      neg_class:
      - app_process_years_2013_2015
      example_hypotheses:
      - talks more about online applications
      - mentions in person interview
    pre_post_recession:
      pos_desc: are job postings from 2004 to 2006
      neg_desc: are job postings from 2007 to 2009
      pos_class:
      - job_req_years_2004_2007
      neg_class:
      - job_req_years_2007_2010
      example_hypotheses:
      - requires more experience
      - lists specific languages
      - requires a Master's degree
    reqs_pre_post_2012:
      pos_desc: are job postings from 2010 to 2012
      neg_desc: are job postings from 2013 to 2014
      pos_class:
      - job_req_years_2010_2013
      neg_class:
      - job_req_years_2013_2015
      example_hypotheses:
      - requires newer programming languages
      - requires more education
  flip: true
- generation: the type of role offered
  applications:
  - target: the expectations and responsibilities of each role, such as specific skills
      or experiences
    user: a job seeker figuring out which role is right for me
    example_hypotheses:
    - emphasizes the importance of company values and ethics
    - requires familiarity with test automation processes
    - Requires travelling for extended periods of time
    v2-origid: 74
    purely_exploratory: false
  pair_type: subject
  pairs:
    junior_senior:
      pos_desc: are job postings for junior positions
      neg_desc: are job postings for senior positions
      pos_class:
      - job_req_senior_sw_dev
      - job_req_senior_sw_eng
      - job_req_senior_qa_eng
      neg_class:
      - job_req_qa_eng
      - job_req_sw_eng
      - job_req_sw_dev
      example_hypotheses:
      - talks about senior positions
      - mentions leadership experience
    sw_qa:
      pos_desc: are job postings for software positions
      neg_desc: are job postings for quality assurance positions
      pos_class:
      - job_desc_sw_dev
      - job_desc_senior_sw_dev
      - job_desc_sw_eng
      - job_desc_senior_sw_eng
      neg_class:
      - job_desc_qa_eng
      - job_desc_senior_qa_eng
      example_hypotheses:
      - talks less about testing
      - talks more about development
      - mentions Java
  flip: true
blm_countermovements:
- generation: the hashtags included in the Tweet
  applications:
  - target: the arguments made by each movement
    user: a sociologist studying cultural movements
    example_hypotheses:
    - includes criticism of media coverage and representation
    - mentions the success of peaceful protests
    - emphasizes freedom of speech for protesters
    - mentions the safety benefits of police officers
    v2-origid: 76
    purely_exploratory: false
  - target: the attitude of each movement
    user: a sociologist studying cultural movements
    example_hypotheses:
    - mentions the importance of respecting all life
    - uses language advocating for unity and solidarity
    - uses language of acceptance, understanding, and respect
    - expresses a desire for change and reform in the justice system
    - contains language that promotes empathy towards other races
    - uses language that acknowledges racism
    v2-origid: 2076
    purely_exploratory: false
  pair_type: subject
  pairs:
    all_white:
      pos_desc: 'are Tweets containing #AllLivesMatter'
      neg_desc: 'are Tweets in support #WhiteLivesMatter'
      pos_class:
      - all_lives_matter
      neg_class:
      - white_lives_matter
      example_hypotheses:
      - does not demean black people
      - does not mention white people
      - is moderate in tone
    blue_white:
      pos_desc: 'are Tweets in support of #BlueLivesMatter'
      neg_desc: 'are Tweets in support of #WhiteLivesMatter'
      pos_class:
      - blue_lives_matter
      neg_class:
      - white_lives_matter
      example_hypotheses:
      - mentions support for police
      - does not demean black people
      - mentions crime
    blue_all:
      pos_desc: 'are Tweets in support of #BlueLivesMatter'
      neg_desc: 'are Tweets in support of #AllLivesMatter'
      pos_class:
      - blue_lives_matter
      neg_class:
      - all_lives_matter
      example_hypotheses:
      - mentions support for police
      - talks about crime
      - is not outwardly racist
  flip: true
cah:
- generation: how many times the card won
  applications:
  - target: which topics tend to be funnier
    user: a rhetoric researcher studying humor
    example_hypotheses:
    - is about intercourse
    - mentions death and violence
    - brings up romantic relationships
    purely_exploratory: false
  - target: what kinds of punchlines are funny
    user:  a rhetoric researcher studying humor
    example_hypotheses:
    - are highly related to the setup
    - are short
    - include a play on words
    purely_exploratory: false
  pair_type: judgement
  pairs:
    white_cards:
      pos_desc: are white cards that won the most
      neg_desc: are white cards that won the least
      pos_class:
      - cards_funny
      neg_class:
      - cards_not_funny
    jokes_funny_somewhat_funny:
      pos_desc: are black card-white card pairs that won very often
      neg_desc: are black card-white card pairs that won somewhat often
      pos_class:
      - jokes_very_funny
      neg_class:
      - jokes_somewhat_funny
    jokes_funny_unfunny:
      pos_desc: are black card-white card pairs that won often
      neg_desc: are black card-white card pairs that usually lost
      pos_class:
      - jokes_very_funny
      - jokes_funny
      neg_class:
      - jokes_not_funny
  flip: true
clickbait_headlines:
- generation: the year it was published
  applications:
  - target: which specific topics dominate the news from year to year
    user: a researcher studying misinformation
    example_hypotheses:
    - mentions gun control or gun related issues
    - mentions the death of a celebrity
    - mentions natural disasters and weather events
    v2-origid: 79
    purely_exploratory: false
  pair_type: time
  pairs:
    pre_post_2013:
      pos_desc: are clickbait headlines from 2010 to 2012
      neg_desc: are clickbait headlines from 2013 to 2015
      pos_class:
      - '2010'
      - '2011'
      - '2012'
      neg_class:
      - '2013'
      - '2014'
      - '2015'
      example_hypotheses:
      - talks more about Mitt Romney
      - talks more about Obama
  flip: true
convincing_arguments:
- generation: how convincing annotators judged the arguments
  applications:
  - target: the rhetorical devices used by convincing arguments
    user: a student writing an argumentative essay who hopes to improve their writing
    example_hypotheses:
    - uses personal stories and anecdotes to bring the argument to life
    - uses vivid language to paint a picture of the situation
    - acknowledges other perspectives, but provides reasons for why their own is more
      valid
    v2-origid: 80
    purely_exploratory: false
  pair_type: judgement
  pairs:
    convincing_unconvincing:
      pos_desc: are convincing arguments
      neg_desc: are unconvincing arguments
      pos_class:
      - somewhat_convincing
      - convincing
      neg_class:
      - unconvincing
      example_hypotheses:
      - provides more explanation
      - gives examples
      - does not insult others
      - cites evidence
    convincing_somewhat_convincing:
      pos_desc: are very convincing arguments
      neg_desc: are somewhat convincing arguments
      pos_class:
      - convincing
      neg_class:
      - somewhat_convincing
      example_hypotheses:
      - provides a clear explanation
      - links a sources
      - is confident
  flip: true
craigslist_negotiations:
- generation: the price of the good being sold
  applications:
  - target: the speaking style of negotiators with different stakes
    user: a business professor interested in negotiation styles
    example_hypotheses:
    - uses more sophisticated vocabulary
    - uses more metaphors to describe the car
    - uses more bargaining-related words, such as 'meet', 'budge' and 'deal'
    v2-origid: 83
    purely_exploratory: false
  pair_type: subject
  pairs:
    bike_price:
      pos_desc: are Craigslist negotiations for expensive bikes
      neg_desc: are Craigslist negotiations for cheaper bikes
      pos_class:
      - bike_high
      neg_class:
      - bike_low
      example_hypotheses:
      - is more polite
      - offers higher prices
    car_price:
      pos_desc: are Craigslist negotiations for expensive cars
      neg_desc: are Craigslist negotiations for cheaper cars
      pos_class:
      - car_high
      neg_class:
      - car_low
      example_hypotheses:
      - is more polite
      - offers higher prices
    housing_price:
      pos_desc: are Craigslist negotiations for expensive housing
      neg_desc: are Craigslist negotiations for cheaper housing
      pos_class:
      - housing_high
      neg_class:
      - housing_low
      example_hypotheses:
      - is more polite
      - offers higher prices
  flip: true
- generation: whether a transaction eventually occured
  applications:
  - target: the speaking style of successful conversations
    user: a Craigslist customer who wants to negotiate well
    example_hypotheses:
    - uses polite language and shows respect for the seller's offer
    - contains words of appreciation
    - uses positive language
    v2-origid: 85
    purely_exploratory: false
  pair_type: judgement
  pairs:
    success_failure:
      pos_desc: are succesful Craigslist negotiations
      neg_desc: are unsuccessful Craigslist negotiations
      pos_class:
      - success
      neg_class:
      - failure
      example_hypotheses:
      - is more cordial
      - offers higher prices
      - outline a long conversation
      - expresses appreciation
  flip: true
dice_jobs:
- generation: the company offering the position
  applications:
  - target: what specific skills different companies require
    user: a recent STEM graduate looking for suitable jobs
    example_hypotheses:
    - refers to network security, Information Assurance, and DISA STIGs
    - requires knowledge of JUNOS and other industry vendor certifications such as
      JNCIA, JNCIS, JNCIP, JNCIE and CCNA, CCNP
    - mentions familiarity with SEI CMM/CMMI and military or IEEE software engineering
      practices
    v2-origid: 88
    purely_exploratory: false
  pair_type: authorship
  pairs:
    amazon_dell:
      pos_desc: are job postings for Amazon
      neg_desc: are job postings for Dell
      pos_class:
      - amazon
      neg_class:
      - dell
      example_hypotheses:
      - talks more about Kotlin, a programming language
      - talks more about Java, a programming language
      - talks more about leadership principles
    jpm_chase_deloitte:
      pos_desc: are job postings for JP Morgan Chase
      neg_desc: are job postings for Deloitte
      pos_class:
      - jpm
      neg_class:
      - deloitte
      example_hypotheses:
      - talks more about banking
      - mentions finance background
      - talks less about consulting
    northup_grumman_leidos:
      pos_desc: are job postings for Northup Grumman
      neg_desc: are job postings for Leidos
      pos_class:
      - northup_grumman
      neg_class:
      - leidos
      example_hypotheses:
      - talks more about aerospace
      - mentions weapons
  flip: true
diplomacy_deception:
- generation: whether the players were telling the truth
  applications:
  - target: the speaking style of liars, so I can tell who might be lying
    user: a sociologist studying lying in games
    example_hypotheses:
    - uses words such as 'really' or 'of course' to emphasize a point
    - uses more questions than statements
    - uses a cooperative tone to establish trust and cooperation
    v2-origid: 89
    purely_exploratory: false
  pair_type: intent
  pairs:
    truth_lie:
      pos_desc: are true statements in a game
      neg_desc: are deceptive statements in a game
      pos_class:
      - truth
      neg_class:
      - lie
      example_hypotheses:
      - makes fewer requests
      - praises the other player less
      - is more straightforward
  flip: true
drug_experiences:
- generation: the substance used during the experience
  applications:
  - target: the specific bodily experiences caused by each drug
    user: a medical researcher researching effects of drugs
    example_hypotheses:
    - mentions feeling as if time is distorted
    - mentions feeling like sinking into the environment
    - mentions feelings of ego surrender or understanding of interconnectedness of
      all things
    v2-origid: 92
    purely_exploratory: false
  - target: which drug is more dangerous to the user
    user: a government regulator considering legalizing a drug
    example_hypotheses:
    - mentions feeling extreme paranoia
    - mentions feeling manipulated by the drug
    - describes throwing up
    v2-origid: 2092
    purely_exploratory: false
  pair_type: subject
  pairs:
    cocaine_molly:
      pos_desc: are accounts of cocaine use
      neg_desc: are accounts of using MDMA/molly
      pos_class:
      - cocaine
      neg_class:
      - mdma
      example_hypotheses:
      - talks about energy
      - does not mention concerts
    lsd_dxm:
      pos_desc: are accounts of LSD (a psychedelic) use
      neg_desc: are accounts of DXM (cough syrup) use
      pos_class:
      - lsd
      neg_class:
      - dxm
      example_hypotheses:
      - more realistic experiences
      - talks more about visions
    lsd_shrooms:
      pos_desc: are accounts of LSD (a psychedelic) use
      neg_desc: are accounts of mushroom use
      pos_class:
      - lsd
      neg_class:
      - mushrooms
      example_hypotheses:
      - talks about higher energy
      - mentions fewer hallucinations
    shrooms_dxm:
      pos_desc: are accounts of mushroom use
      neg_desc: are accounts of DXM (cough syrup) use
      pos_class:
      - mushrooms
      neg_class:
      - dxm
      example_hypotheses:
      - mentions stomach pain
  flip: true
echr_decisions:
- generation: the ruling of the court
  applications:
  - target: what kinds of evidence convince the court that there is a violation
    user: a lawyer planning a defense for my defendant
    example_hypotheses:
    - brings up a voice recording
    - refers to witnesses or expert testimony
    - talks about DNA evidence
    v2-origid: 94
    purely_exploratory: false
  - target: which actions are more likely to constitute a violation
    user: a legal scholar studying court ideology
    example_hypotheses:
    - mentions a lack of medical treatment or inadequate medical care
    - describes a case involving a minor or children
    - refers to a breach of a human right or fundamental freedom
    v2-origid: 2094
    purely_exploratory: false
  pair_type: judgement
  pairs:
    yes_no_violation:
      pos_desc: are human rights trials where a violation was found
      neg_desc: are human rights trials where no violation was found
      pos_class:
      - violation
      neg_class:
      - no_violation
      example_hypotheses:
      - talks more about witnesses
      - mentions more violence
      - talks about discrimination
      - does not mention intent
  flip: true
essay_scoring:
- generation: grades assigned by readers
  applications:
  - target: the general style of writing readers look for
    user: a student writing an academic paper
    example_hypotheses:
    - uses metaphors and similes to create vivid imagery
    - uses active verbs for a more dynamic writing style
    - uses a lot of alliteration
    v2-origid: 95
    purely_exploratory: false
  pair_type: judgement
  pairs:
    good_bad:
      pos_desc: are essays with good scores
      neg_desc: are essays with bad scores
      pos_class:
      - good_essays
      neg_class:
      - bad_essays
      example_hypotheses:
      - uses more sophisticated words
      - is more coherent
      - has fewer spelling mistakes
      - cites examples
  flip: false
fake_news:
- generation: whether they are legitimate or fake news articles
  applications:
  - target: the writing style of fake news sources
    user: a content moderator looking to flag fake news
    example_hypotheses:
    - uses vague language to describe events or outcomes
    - uses strong emotion-laden words, such as 'furious', 'victimized' and 'horrifying'
    - uses sarcasm and mockery
    v2-origid: 96
    purely_exploratory: false
  - target: the topics discussed by fake news sources
    user: a content moderator looking to flag fake news
    example_hypotheses:
    - mentions Russia or Ukraine
    - contains allegations against well-known figures without evidence
    - talks about politicians and voting
    v2-origid: 2096
    purely_exploratory: false
  pair_type: intent
  pairs:
    fake_legit:
      pos_desc: are fake news articles
      neg_desc: are legitimate news articles
      pos_class:
      - fake
      neg_class:
      - legit
      example_hypotheses:
      - uses extreme language
      - cites fewer sources
      - is more political
      - is fearful
  flip: true
fomc_speeches:
- generation: who gave the speech
  applications:
  - target: the economic ideology of different Fed chairpeople
    user: a economist studying the stances of Fed board members
    example_hypotheses:
    - mentions the need for financial modernization
    - mentions the benefits of bank regulation
    - uses language emphasizing the importance of small businesses
    v2-origid: 100
    purely_exploratory: false
  - target: the policy priorities of different Fed chairpeople
    user: a economist studying the stances of Fed board members
    example_hypotheses:
    - discusses the Federal Reserve's role in banking supervision and regulation
    - mentions the Gramm-Leach-Bliley Act
    - discusses the need for convergence of regulatory standards
    v2-origid: 2100
    purely_exploratory: false
  pair_type: authorship
  pairs:
    greenspan_bernanke:
      pos_desc: are FOMC speeches from Chairman Greenspan
      neg_desc: are FOMC speeches from Chairman Bernanke
      pos_class:
      - greenspan_speeches
      neg_class:
      - bernanke_speeches
      example_hypotheses:
      - calls for more action
      - talks more about growth
      - talks less about targets
      - mentions the recession
    yellen_powell:
      pos_desc: are FOMC speeches from Chairman Yellen
      neg_desc: are FOMC speeches from Chairman Powell
      pos_class:
      - yellen_speeches
      neg_class:
      - powell_chair_speeches
      example_hypotheses:
      - does not mention inflation
      - talks about unemployment
      - is optimistic
    powell_gov_chair:
      pos_desc: are FOMC speeches from Jerome Powell as Fed chariman
      neg_desc: are FOMC speeches from Jerome Powell as a governor
      pos_class:
      - powell_chair_speeches
      neg_class:
      - powell_gov_speeches
      example_hypotheses:
      - mentions quantitative easing
      - support financial regulation
      - is more assertive
    meyer_ferguson:
      pos_desc: are FOMC speeches from Fed Governor Meyer
      neg_desc: are FOMC speeches from Fed Vice Chairman Ferguson
      pos_class:
      - meyer_speeches
      neg_class:
      - ferguson_speeches
      example_hypotheses:
      - pushes to raise interest rates
      - is more aggressive
  flip: true
- generation: the year the speech was given
  applications:
  - target: the trends in Fed policy and priorities over the years
    user: a economic historian studying trends in Fed policy
    example_hypotheses:
    - mentions the importance of transparency and accountability for monetary policymaking
    - mentions the importance of increasing representation of black women in economics
    - references the 2 percent inflation objective
    v2-origid: 102
    purely_exploratory: false
  pair_type: time
  pairs:
    greenspan_bernanke_year:
      pos_desc: are FOMC speeches from before 2006
      neg_desc: are FOMC speeches from 2006 to 2014
      pos_class:
      - greenspan_years
      neg_class:
      - bernanke_years
      example_hypotheses:
      - calls for more action
      - talks more about growth
      - talks less about targets
    yellen_powell_years:
      pos_desc: are FOMC speeches from 2014 to 2018
      neg_desc: are FOMC speeches from after 2018
      pos_class:
      - yellen_years
      neg_class:
      - powell_years
      example_hypotheses:
      - talks more about unemployment
      - does not mention COVID-19
      - talks about economic growth
  flip: true
- generation: the state of the economy during the speech
  applications:
  - target: what outcomes the Fed prioritizes under different economic conditions
    user: an economist studying patterns of Fed behavior
    example_hypotheses:
    - focuses on poverty and inequality
    - discusses how to stimulate the economy and create jobs
    - discusses strategies to achieve maximum employment and price stability
    v2-origid: 105
    purely_exploratory: false
  pair_type: socioeconomic
  pairs:
    growth:
      pos_desc: are FOMC speeches during periods of high GDP growth
      neg_desc: are FOMC speeches during periods of low GDP growth
      pos_class:
      - high_growth
      neg_class:
      - low_growth
      example_hypotheses:
      - talks more about inflation
      - talks about a strong economy
    interest_rate:
      pos_desc: are FOMC speeches during periods of high interest rates
      neg_desc: are FOMC speeches during periods of low interest rates
      pos_class:
      - high_ir
      neg_class:
      - low_ir
      example_hypotheses:
      - talks more about inflation
      - worries more about recession
      - talks about job losses
    unemployment:
      pos_desc: are FOMC speeches during periods of high unemployment
      neg_desc: are FOMC speeches during periods of low unemployment
      pos_class:
      - high_unemp
      neg_class:
      - low_unemp
      example_hypotheses:
      - talks more about unemployment
      - talks about lowering the interest rate
      - talks less about inflation
      - mentions popular opinion
  flip: true
genius_lyrics:
- generation: the year the song was released
  applications:
  - target: the specific topics of music throughout the years
    user: a record label looking for the next big hit
    example_hypotheses:
    - references religious topics such as God or prayer
    - mentions wealth and affluence, such as money or riches
    - mentions a sense of belonging, such as a crew or a team
    v2-origid: 107
    purely_exploratory: false
  - target: the artistic style of music throughout the years
    user: a record label looking for the next big hit
    example_hypotheses:
    - uses a lot of imagery in the lyrics
    - uses a lot of repetition in the lyrics
    - is written in a conversational style, i.e., two people talking
    v2-origid: 2107
    purely_exploratory: false
  pair_type: time
  pairs:
    2ks_2010s:
      pos_desc: are 2000-2010 song lyrics
      neg_desc: are 2010s song lyrics
      pos_class:
      - '2000'
      - '2001'
      - '2002'
      - '2003'
      - '2004'
      - '2005'
      - '2006'
      - '2007'
      - '2008'
      - '2009'
      neg_class:
      - '2010'
      - '2011'
      - '2012'
      - '2013'
      - '2014'
      - '2015'
      - '2016'
      - '2017'
      - '2018'
      - '2019'
      example_hypotheses:
      - does not mention technology
      - is more optimistic
    70s_80s:
      pos_desc: are seventies song lyrics
      neg_desc: are eighties song lyrics
      pos_class:
      - '1970'
      - '1971'
      - '1972'
      - '1973'
      - '1974'
      - '1975'
      - '1976'
      - '1977'
      - '1978'
      - '1979'
      neg_class:
      - '1980'
      - '1981'
      - '1982'
      - '1983'
      - '1984'
      - '1985'
      - '1986'
      - '1987'
      - '1988'
      - '1989'
      example_hypotheses:
      - is less passionate
      - talks about rebellion
    80s_90s:
      pos_desc: are eighties song lyrics
      neg_desc: are nineties song lyrics
      pos_class:
      - '1980'
      - '1981'
      - '1982'
      - '1983'
      - '1984'
      - '1985'
      - '1986'
      - '1987'
      - '1988'
      - '1989'
      neg_class:
      - '1990'
      - '1991'
      - '1992'
      - '1993'
      - '1994'
      - '1995'
      - '1996'
      - '1997'
      - '1998'
      - '1999'
      example_hypotheses:
      - is less passionate
      - mentions the government
    90s_2ks:
      pos_desc: are nineties song lyrics
      neg_desc: are 2000-2010 song lyrics
      pos_class:
      - '1990'
      - '1991'
      - '1992'
      - '1993'
      - '1994'
      - '1995'
      - '1996'
      - '1997'
      - '1998'
      - '1999'
      neg_class:
      - '2000'
      - '2001'
      - '2002'
      - '2003'
      - '2004'
      - '2005'
      - '2006'
      - '2007'
      - '2008'
      - '2009'
      example_hypotheses:
      - has hip-hop lyrics
      - mentions rebellion
      - uses profanity
  flip: true
- generation: the genre of the lyrics
  applications:
  - target: the topics that define each genre
    user: an artist trying to figure out which genre my music belongs to
    example_hypotheses:
    - mentions relationships and heartbreak
    - references to drugs and alcohol
    - mentions struggles, including poverty and racism
    v2-origid: 113
    purely_exploratory: false
  - target: the artistic style that defines each genre
    user: an artist trying to figure out which genre my music belongs to
    example_hypotheses:
    - uses aggressive language and threats
    - uses alliteration or slant rhyme in the lyrics
    - uses metaphors and similes
    v2-origid: 2113
    purely_exploratory: false
  pair_type: subject
  pairs:
    alt_pop:
      pos_desc: are from the alternative genre
      neg_desc: are from the pop genre
      pos_class:
      - alt
      neg_class:
      - pop
      example_hypotheses:
      - discusses personal issues
      - does not rhyme
    alt_rock:
      pos_desc: are from the alternative rock genre
      neg_desc: are from the rock genre
      pos_class:
      - alt_rock
      neg_class:
      - rock
      example_hypotheses:
      - discusses personal issues
      - talks about relationships
    r_b_soul:
      pos_desc: are from the R&B genre
      neg_desc: are from the soul genre
      pos_class:
      - r_b
      neg_class:
      - soul
      example_hypotheses:
      - is not religious
      - rhymes
    rap_trap:
      pos_desc: are from the rap genre
      neg_desc: are from the trap genre
      pos_class:
      - rap
      neg_class:
      - trap
      example_hypotheses:
      - uses longer sentences
      - does not repeat
      - talks about personal issues
  flip: true
- generation: the artist
  applications:
  - target: the specific topics of each musician
    user: a music fanatic hoping to understand each artist
    example_hypotheses:
    - speaks of potential, ambition, and success
    - explores themes of loyalty, betrayal and redemption
    - mentions racism and police brutality
    - mentions religious faith and beliefs
    v2-origid: 117
    purely_exploratory: false
  - target: the artistic style of each musician
    user: a music fanatic hoping to understand each artist
    example_hypotheses:
    - uses street slang and references to Compton
    - uses metaphorical language to describe their experiences
    - features a recurring chorus
    v2-origid: 2117
    purely_exploratory: false
  pair_type: authorship
  pairs:
    ariana_grande_beyonce:
      pos_desc: are Ariana Grande song lyrics
      neg_desc: are Beyonce song lyrics
      pos_class:
      - ariana_grande
      neg_class:
      - beyonce
      example_hypotheses:
      - talks more about relationships
      - is less empowering
    drake_kanye:
      pos_desc: are Drake song lyrics
      neg_desc: are Kanye song lyrics
      pos_class:
      - drake
      neg_class:
      - kanye
      example_hypotheses:
      - talks less about God
      - talks more about money
      - brags about wealth
    eminem_logic:
      pos_desc: are Eminem song lyrics
      neg_desc: are Logic song lyrics
      pos_class:
      - eminem
      neg_class:
      - logic
      example_hypotheses:
      - use clever rhymes
      - is less repetitive
      - has longer sentences
    kendrick_j_cole:
      pos_desc: are Kendrick Lamar song lyrics
      neg_desc: are J. Cole song lyrics
      pos_class:
      - kendrick
      neg_class:
      - j_cole
      example_hypotheses:
      - talks more about money
      - uses more profanity
      - demeans women
    weeknd_post_malone:
      pos_desc: are The Weeknd song lyrics
      neg_desc: are Post Malone song lyrics
      pos_class:
      - ariana_grande
      neg_class:
      - beyonce
      example_hypotheses:
      - talks more about relationships
  flip: true
- generation: where the song was created
  applications:
  - target: the topics that come up in music from different places
    user: a music researcher studying the emergence of music in different places
    example_hypotheses:
    - describes feelings of hopelessness or sadness
    - uses words associated with traveling and exploration
    - mentions London and other UK cities
    v2-origid: 120
    purely_exploratory: false
  - target: the artistic style of music from different places
    user: a music researcher studying the emergence of music in different places
    example_hypotheses:
    - uses religious imagery or references to spirituality
    - uses strong language to express feelings of anger and frustration
    - utilizes slang and colloquial language
    v2-origid: 2120
    purely_exploratory: false
  pair_type: location
  pairs:
    east_west_coast:
      pos_desc: are east coast song lyrics
      neg_desc: are west coast song lyrics
      pos_class:
      - west_coast
      neg_class:
      - east_coast
      example_hypotheses:
      - talks about sunshine
      - mentions Los Angeles
      - mentions California
    uk_france:
      pos_desc: are UK lyrics
      neg_desc: are French lyrics
      pos_class:
      - uk
      neg_class:
      - france
      example_hypotheses:
      - mentions everyday struggles
      - is not about relationships
  flip: true
- generation: how many views the song received on Genius
  applications:
  - target: the topics that a pop audience looks for in a song
    user: a record label looking for the next big hit
    example_hypotheses:
    - mentions drugs, alcohol, and partying
    - describes a journey or travel
    - mentions controversial topics such as drugs or criminal activities
    v2-origid: 121
    purely_exploratory: false
  - target: the artistic style that a pop audience looks for in a song
    user: a record label looking for the next big hit
    example_hypotheses:
    - has a lot of energy, with a fast tempo
    - uses repetitive phrases in the chorus
    - contains lots of braggadocio
    v2-origid: 2121
    purely_exploratory: false
  pair_type: judgement
  pairs:
    viewcount:
      pos_desc: are highly viewed song lyrics
      neg_desc: are song lyrics with moderate views
      pos_class:
      - high_views
      neg_class:
      - mid_views
      example_hypotheses:
      - talks about personal topics
      - has a catchy rhyme
  flip: true
happy_moments:
- generation: the age of the respondent
  applications:
  - target: what specific experiences make us happy as we grow older
    user: a psychologist studying the effect of aging on happiness
    example_hypotheses:
    - involves doing physical activities, such as going to the park, working out,
      or taking a walk
    - references reunions with old friends
    - references psychedelic experiences
    v2-origid: 122
    purely_exploratory: false
  pair_type: demographic
  pairs:
    college_20s:
      pos_desc: include self-reported happy moments from 18-21 year olds
      neg_desc: include self-reported happy moments from 22-25 year olds
      pos_class:
      - 18-21
      neg_class:
      - 22-25
      example_hypotheses:
      - talks about traveling
      - talks about new places
      - talks about relationships
      - mentions college
    early_late_20s:
      pos_desc: include self-reported happy moments from 22-25 year olds
      neg_desc: include self-reported happy moments from 26-35 year olds
      pos_class:
      - 22-25
      neg_class:
      - 26-35
      example_hypotheses:
      - talks about traveling
      - talks about new places
      - mentions work
    late_20s_30s:
      pos_desc: include self-reported happy moments from 26-35 year olds
      neg_desc: include self-reported happy moments from 36-45 year olds
      pos_class:
      - 26-35
      neg_class:
      - 36-45
      example_hypotheses:
      - mentions marraige
      - does not mention children
    30s_40s:
      pos_desc: include self-reported happy moments from 36-45 year olds
      neg_desc: include self-reported happy moments from 46+ year olds
      pos_class:
      - 36-45
      neg_class:
      - 46+
      example_hypotheses:
      - talks about children
      - talks more about work
  flip: true
- generation: the type of happiness experienced (achivement, affection, bonding, enjoying
    the moment, exercise, leisure, and nature) from crowdworker annotations
  applications:
  - target: which part of each experience makes people happy
    user: a psychologist trying to precisely define types of emotions
    example_hypotheses:
    - refers to a moment of joy with an extended family member
    - involves physical contact or expressions of physical affection
    - mentions events that involve receiving gifts or presents
    v2-origid: 126
    purely_exploratory: false
  pair_type: subject
  pairs:
    affection_bonding:
      pos_desc: include self-reported happy moments about affection
      neg_desc: include self-reported happy moments about bonding
      pos_class:
      - affection
      neg_class:
      - bonding
      example_hypotheses:
      - talks more about relatives
      - talk more about love
    enjoying_leisure:
      pos_desc: include self-reported happy moments about enjoying the moment
      neg_desc: include self-reported happy moments about leisure
      pos_class:
      - enjoy_the_moment
      neg_class:
      - leisure
      example_hypotheses:
      - talks more about experiences
      - talks less about relaxing
  flip: true
- generation: the gender or familial status of the respondent
  applications:
  - target: how interpersonal relationships shape happiness
    user: a sociologist studying intimate relationships
    example_hypotheses:
    - involve small acts of self-care or indulgence
    - mentions activities related to home maintenance or renovation
    - mentions successful experiences, such as getting a job or a good deal
    v2-origid: 131
    purely_exploratory: false
  pair_type: demographic
  pairs:
    gender:
      pos_desc: include self-reported happy moments from males
      neg_desc: include self-reported happy moments from females
      pos_class:
      - male
      neg_class:
      - female
      example_hypotheses:
      - talks about work
      - talks about friends
      - mentions sports
    marraige:
      pos_desc: include self-reported happy moments from unmarried people
      neg_desc: include self-reported happy moments from married people
      pos_class:
      - single
      neg_class:
      - married
      example_hypotheses:
      - talks about dating
      - talks about relatives
      - mentions friends
    parenthood:
      pos_desc: include self-reported happy moments from parents
      neg_desc: include self-reported happy moments from non-parents
      pos_class:
      - parent
      neg_class:
      - not_parent
      example_hypotheses:
      - talks about children
      - talks about spouses
      - mentions a baby
    separation_divorce:
      pos_desc: include self-reported happy moments from people who have separated
        (without divorcing)
      neg_desc: include self-reported happy moments from divorced people
      pos_class:
      - separated
      neg_class:
      - divorced
      example_hypotheses:
      - talks about seeing children
      - mentions friends
      - does not mention dates
  flip: true
- generation: the respondent's nationality
  applications:
  - target: what specific experiences or topics make people happy across the world
    user: a demographer comparing the values of different cultures
    example_hypotheses:
    - involves physical accomplishments, such as running or hitting a record time
      on a mile
    - mentions material possessions, such as cars and phones
    - mention activities related to the church
    v2-origid: 134
    purely_exploratory: false
  pair_type: location
  pairs:
    usa_canada:
      pos_desc: include self-reported happy moments from the United States
      neg_desc: include self-reported happy moments from Canada
      pos_class:
      - usa
      neg_class:
      - canada
      example_hypotheses:
      - talks about friends
      - does not discuss nature
      - talks about consumption
      - mentions sports
    usa_india:
      pos_desc: include self-reported happy moments from the United States
      neg_desc: include self-reported happy moments from India
      pos_class:
      - usa
      neg_class:
      - india
      example_hypotheses:
      - talks about friends
      - talks about leisure
      - talks less about others
      - does not mention family
    usa_venezuela:
      pos_desc: include self-reported happy moments from the United States
      neg_desc: include self-reported happy moments from Venezuela
      pos_class:
      - usa
      neg_class:
      - venezuela
      example_hypotheses:
      - does not mention money
      - mentions relationships
      - does not talk about nature
  flip: true
immigration_speeches:
- generation: who gave the speech
  applications:
  - target: the specific beliefs of different politicians on immigration
    user: a political scientist studying stances on immigration
    example_hypotheses:
    - mentions the need to clean up the Social Security database and Immigration Service
      database
    - argues that the SAW Program should be utilized to address labor shortages
    - mentions the need for the President to report to Congress on the cost of State
      incarceration of illegal aliens
    v2-origid: 147
    purely_exploratory: false
  - target: the stereotypes and metaphors different politicians appeal to
    user: a political scientist studying stances on immigration
    example_hypotheses:
    - mentions Dreamers and their plight
    - calls for protection of children from deportation
    - mentions the high dropout rate among Hispanic/Latino students
    v2-origid: 148
    purely_exploratory: false
  pair_type: authorship
  pairs:
    trump_obama:
      pos_desc: are speeches given by President Trump
      neg_desc: are speeches given by President Obama
      pos_class:
      - speaker_trump
      neg_class:
      - speaker_obama
      example_hypotheses:
      - is hostile towards immigrants
      - supports a stronger border
    trump_pence:
      pos_desc: are speeches given by President Trump
      neg_desc: are speeches given by Vice President Pence
      pos_class:
      - speaker_trump
      neg_class:
      - speaker_pence
      example_hypotheses:
      - is outwardly hostile towards immigrants
      - focuses on Mexican immigrants
      - mentions the border
    bush_jr_sr:
      pos_desc: are speeches given by President Bush Jr.
      neg_desc: are speeches given by President Bush Sr.
      pos_class:
      - speaker_bush_jr
      neg_class:
      - speaker_bush_sr
      example_hypotheses:
      - mentions the Middle East
      - talks about terrorism
    bush_sr_reagan:
      pos_desc: are speeches given by President Bush Sr.
      neg_desc: are speeches given by President Reagan
      pos_class:
      - speaker_bush_sr
      neg_class:
      - speaker_reagan
      example_hypotheses:
      - is warmer towards immigrants
      - mentions the need for high-skilled labor
      - talks about Asian immigrants
    clinton_obama:
      pos_desc: are speeches given by President Bush Jr.
      neg_desc: are speeches given by Bill
      pos_class:
      - speaker_clinton
      neg_class:
      - speaker_obama
      example_hypotheses:
      - is hostile towards immigrants
      - talks about border security
    sessions_kennedy:
      pos_desc: are speeches given by Congressperson Jeff Sessions
      neg_desc: are speeches given by Congressperson Edward Kennedy
      pos_class:
      - con_speaker_sessions
      neg_class:
      - con_speaker_kennedy
      example_hypotheses:
      - is hostile towards immigrants
      - mentions the Southern border
      - uses strong language
    durbin_simpson:
      pos_desc: are speeches given by Congressperson Richard Durbin
      neg_desc: are speeches given by Congressperson Mike Simpson
      pos_class:
      - con_speaker_durbin
      neg_class:
      - con_speaker_simpson
      example_hypotheses:
      - mentions the DREAM act
      - is more welcoming towards immigrants
      - supports President Obama
  flip: true
- generation: whether the statement was written or spoken
  applications:
  - target: how the speaking style of politicians changes depending on medium
    user: a political scientist studying stances on immigration
    example_hypotheses:
    - uses more personal anecdotes
    - uses more repetition or rhetorical devices
    - 'uses the phrase ''illegal aliens'' more often '
    v2-origid: 149
    purely_exploratory: false
  pair_type: format
  pairs:
    trump_written_spoken:
      pos_desc: are spoken speeches delivered by President Trump
      neg_desc: are written statements by President Trump
      pos_class:
      - speaker_trump_spoken
      neg_class:
      - speaker_trump_written
      example_hypotheses:
      - is less hostile
      - uses more sophisticated language
      - expresses moderation
    obama_written_spoken:
      pos_desc: are written statements by President Obama
      neg_desc: are spoken speeches delivered by President Obama
      pos_class:
      - speaker_obama_written
      neg_class:
      - speaker_obama_spoken
      example_hypotheses:
      - argues for specific policies
      - is less positive
  flip: true
- generation: when the statement was made
  applications:
  - target: how different events and eras influence the general perception of immigrants
    user: a political scientist studying stances on immigration
    example_hypotheses:
    - emphasizes the importance of providing aid and assistance to Italian immigrants
    - mentions the importance of immigrants in the military
    - references the opinion of religious leaders about immigration laws
    v2-origid: 165
    purely_exploratory: false
  - target: the dominant stereotypes and metaphors of each time era
    user: a political scientist studying stances on immigration
    example_hypotheses:
    - uses language that paints immigrants in a positive light, such as describing
      their contribution to society
    - uses metaphors that describe immigrants as a burden, such as using words like
      'flood' or 'invasion'
    - describes the situation of immigrants in a negative way, or paints them as a
      'problem'
    v2-origid: 162
    purely_exploratory: false
  pair_type: time
  pairs:
    obama_term:
      pos_desc: are statements by President Obama in his second term
      neg_desc: are statements by President Obama in his first term
      pos_class:
      - speaker_obama_01-26-2013_05-07-2017
      neg_class:
      - speaker_obama_01-26-2009_01-25-2013
      example_hypotheses:
      - is more firm on border security
      - is less optimistic
    bush_911:
      pos_desc: are statements by President Bush Jr. after 9/11
      neg_desc: are statements by President Bush Jr. prior to 9/11
      pos_class:
      - speaker_bush_jr_01-20-2001_09-10-2001
      neg_class:
      - speaker_obama_01-26-2009_01-25-2013
      example_hypotheses:
      - talks about terrorism
      - mentions the Middle East
    chinese_early_mid:
      pos_desc: are statements about Chinese immigrants between 1873 and 1934
      neg_desc: are statements about Chinese immigrants between 1935 and 1956
      pos_class:
      - period_early_chinese
      neg_class:
      - period_mid_chinese
      example_hypotheses:
      - talks about foreigners disparagingly
      - mentions aliens
    chinese_late_mid:
      pos_desc: are statements about Chinese immigrants between 1957 and 2020
      neg_desc: are statements about Chinese immigrants between 1935 and 1956
      pos_class:
      - period_late_chinese
      neg_class:
      - period_mid_chinese
      example_hypotheses:
      - talks about national security threats
    mexican_early_mid:
      pos_desc: are statements about Mexican immigrants between 1873 and 1934
      neg_desc: are statements about Mexican immigrants between 1935 and 1956
      pos_class:
      - period_early_mexican
      neg_class:
      - period_mid_mexican
      example_hypotheses:
      - is hostile towards immigrants
      - talks about labor
    mexican_late_mid:
      pos_desc: are statements about Mexican immigrants between 1957 and 2020
      neg_desc: are statements about Mexican immigrants between 1935 and 1956
      pos_class:
      - period_late_mexican
      neg_class:
      - period_mid_mexican
      example_hypotheses:
      - talks about illegal immigration
      - is more hostile towards immigrants
      - mentions crime
    italian_early_mid:
      pos_desc: are statements about Italian immigrants between 1873 and 1934
      neg_desc: are statements about Italian immigrants between 1935 and 1956
      pos_class:
      - period_early_italian
      neg_class:
      - period_mid_italian
      example_hypotheses:
      - mentions crime
      - mentions the mafia
      - is hostile
    italian_late_mid:
      pos_desc: are statements about Italian immigrants between 1957 and 2020
      neg_desc: are statements about Italian immigrants between 1935 and 1956
      pos_class:
      - period_late_italian
      neg_class:
      - period_mid_italian
      example_hypotheses:
      - is warm towards immigrants
      - talks about Europe
      - mentions culture
  flip: true
- generation: whether the speaker is in favor of immigration
  applications:
  - target: what specific justifications people in favor of immigration use
    user: a political scientist studying stances on immigration
    example_hypotheses:
    - mentions equality and justice for all
    - highlights the industriousness of immigrants
    - highlights the unfairness of discrimination against immigrants
    v2-origid: 167
    purely_exploratory: false
  pair_type: sentiment
  pairs:
    early_pro_anti:
      pos_desc: are statements supportive of immigration between 1873 and 1934
      neg_desc: are statements opposed to immigration between 1873 and 1934
      pos_class:
      - period_early_pro
      neg_class:
      - period_early_anti
      example_hypotheses:
      - mentions economic benefits
      - talks about labor
    mid_pro_anti:
      pos_desc: are statements supportive of immigration between 1935 and 1956
      neg_desc: are statements opposed to immigration between 1935 and 1956
      pos_class:
      - period_mid_pro
      neg_class:
      - period_mid_anti
      example_hypotheses:
      - mentions humanitarian needs
      - expresses empathy
      - mentions World War II refugees
    late_pro_anti:
      pos_desc: are statements supportive of immigration between 1957 and 2020
      neg_desc: are statements opposed to immigration between 1957 and 2020
      pos_class:
      - period_late_pro
      neg_class:
      - period_late_anti
      example_hypotheses:
      - mentions high-skilled workers
      - talks about H1B visas
      - mentions refugees
  flip: true
- generation: the political party of the speaker
  applications:
  - target: the specific policy priorities of each political party
    user: a political scientist studying stances on immigration
    example_hypotheses:
    - mentions the need for a wall at the border
    - mentions the need to record who is entering the United States
    - calls for better background checks and welfare checks for immigrants
    v2-origid: 172
    purely_exploratory: false
  pair_type: demographic
  pairs:
    early_dem_rep:
      pos_desc: are statements by Democrats on immigration between 1873 and 1934
      neg_desc: are statements by Republicans on immigration between 1873 and 1934
      pos_class:
      - period_early_democrat
      neg_class:
      - period_early_republican
      example_hypotheses:
      - mentions economic benefits
      - talks about labor
    mid_dem_rep:
      pos_desc: are statements by Democrats on immigration between 1935 and 1956
      neg_desc: are statements by Republicans on immigration between 1935 and 1956
      pos_class:
      - period_mid_democrat
      neg_class:
      - period_mid_republican
      example_hypotheses:
      - mentions humanitarian needs
      - mentions World War II
      - talks about labor
    late_dem_rep:
      pos_desc: are statements by Democrats on immigration between 1957 and 2020
      neg_desc: are statements by Republicans on immigration between 1957 and 2020
      pos_class:
      - period_late_democrat
      neg_class:
      - period_late_republican
      example_hypotheses:
      - mentions high-skilled workers
      - mentions humanitarian obligations
      - supports immigration reform
  flip: true
- generation: the geographic region of the speaker's home
  applications:
  - target: how people from different parts of the United States feel about immigrants
    user: a political scientist studying stances on immigration
    example_hypotheses:
    - argues for the importance of creating a safe and orderly flow of refugees
    - discusses efforts to improve the health of migrant workers
    - focuses on the humanitarian aspects of immigration
    v2-origid: 174
    purely_exploratory: false
  pair_type: location
  pairs:
    late_northern_western:
      pos_desc: are statements by Northerners on immigration between 1957 and 2020
      neg_desc: are statements by Westerners on immigration between 1957 and 2020
      pos_class:
      - period_late_northern
      neg_class:
      - period_late_western
      example_hypotheses:
      - is more supportive of immigration
      - does not mention Mexico
      - talks positively about Europeans
    late_northern_southern:
      pos_desc: are statements by Northerners on immigration between 1957 and 2020
      neg_desc: are statements by Southerners on immigration between 1957 and 2020
      pos_class:
      - period_late_northern
      neg_class:
      - period_late_southern
      example_hypotheses:
      - is positive towards immigration
      - does not mention the border
      - talks about refugees
    late_western_southern:
      pos_desc: are statements by Westerners on immigration between 1957 and 2020
      neg_desc: are statements by Southerners on immigration between 1957 and 2020
      pos_class:
      - period_late_western
      neg_class:
      - period_late_southern
      example_hypotheses:
      - are more warm towards immigration
      - mentions farm labor
  flip: true
kickstarter:
- generation: whether the fundraiser succeeded or failed
  applications:
  - target: what kinds of companies succeed
    user: an angel investor
    example_hypotheses:
    - focuses on technology
    - addresses a social or environmental issue
    - mentions a specific target demographic or market
    - uses vivid language to evoke emotion
    - uses a catchy title that draws people in
    - mentions a specific target demographic or market
    - uses words associated with success, such as 'premium', 'revolutionary', or 'innovative'
    - uses emotionally charged words
    - has a clear goal and timeline for completion
    v2-origid: 176
    purely_exploratory: false
  - target: how successful fundraisers market themselves
    user: an angel investor
    example_hypotheses:
    - focuses on technology
    - addresses a social or environmental issue
    - mentions a specific target demographic or market
    - uses vivid language to evoke emotion
    - uses a catchy title that draws people in
    - mentions a specific target demographic or market
    - uses words associated with success, such as 'premium', 'revolutionary', or 'innovative'
    - uses emotionally charged words
    - has a clear goal and timeline for completion
    v2-origid: 176
    purely_exploratory: false
  pair_type: judgement
  pairs:
    state_successful_failed:
      pos_desc: are names of startups that succeeded
      neg_desc: are names of startups that failed
      pos_class:
      - state_successful
      neg_class:
      - state_failed
      example_hypotheses:
      - is about a creative work
      - is a personal hobby
  flip: false
microedit_humor:
- generation: how funny annotators found the sentences
  applications:
  - target: what types of humor people find funny
    user: a rhetoric researcher studying the nature of humor
    example_hypotheses:
    - uses slang or colloquial language
    - include unexpected elements, such as a twist in the plot
    - include a parody of a popular trend, such as making fun of a celebrity
    v2-origid: 177
    purely_exploratory: false
  - target: which specific writing features make a joke not funny
    user: a rhetoric researcher studying the nature of humor
    example_hypotheses:
    - rely more on wordplay than clever puns
    - involves comparisons of incongruous objects
    - refers to current events or popular culture
    v2-origid: 178
    purely_exploratory: false
  pair_type: judgement
  pairs:
    funny_very_funny:
      pos_desc: are sentences edited to be somewhat funny
      neg_desc: are sentences edited to be very funny
      pos_class:
      - funny
      neg_class:
      - very_funny
      example_hypotheses:
      - is not as surprising surprising
      - uses slapstick humor
      - uses profanity
      - relies on wordplay
    unfunny_funny:
      pos_desc: are edited sentences that are not funny
      neg_desc: are edited sentences that are funny
      pos_class:
      - unfunny
      neg_class:
      - funny
      example_hypotheses:
      - is incoherent
      - has a predictable punchline
      - is confusing
  flip: true
monster_jobs:
- generation: where the job was posted
  applications:
  - target: the requirements of jobs in different cities
    user: a recent graduate trying to figure out which city to look for jobs in
    example_hypotheses:
    - mentions a Master's degree or Ph.D in a related field
    - mentions the need to have experience in a public accounting firm or in industry
    - mentions the need for a valid driver's license
    v2-origid: 181
    purely_exploratory: false
  - target: the benefits of jobs in different cities
    user: a recent graduate trying to figure out which city to look for jobs in
    example_hypotheses:
    - offer work-from-home options
    - mention paid parental leave
    - offers insurance benefits
    v2-origid: 2181
    purely_exploratory: false
  pair_type: location
  pairs:
    atl_tampa:
      pos_desc: are job postings in Atlanta, GA
      neg_desc: are job postings in Tampa, FL
      pos_class:
      - atlanta
      neg_class:
      - tampa
      example_hypotheses:
      - offers blue collar jobs
      - mentions service jobs
      - offers a job at a casino
    ca_tx:
      pos_desc: are job postings in California
      neg_desc: are job postings in Texas
      pos_class:
      - sf
      - la
      - berkeley
      neg_class:
      - dallas
      - houston
      - austin
      example_hypotheses:
      - offers tech related jobs
      - offers more benefits
      - offers greater compensation
    sf_nyc:
      pos_desc: are job postings in New York City
      neg_desc: are job postings in San Francisco
      pos_class:
      - nyc
      neg_class:
      - sf
      example_hypotheses:
      - offers more finance jobs
      - talks more about location
      - talks about consulting
      - does not mention technology
  flip: true
movie_tmdb:
- generation: how successful the movie was
  applications:
  - target: which genres movie-goers seem to like
    user: a movie director planning a new film
    example_hypotheses:
    - includes a supernatural element such as a ghost or time travel
    - is set in a dystopian world
    - contains references to ancient mythology or legends
    v2-origid: 184
    purely_exploratory: false
  - target: what specific plot devices are more popular to movie-goers
    user: a movie director planning a new film
    example_hypotheses:
    - uses humor to lighten the mood
    - involves a heroic struggle against a powerful enemy
    - has a romantic subplot
    v2-origid: 187
    purely_exploratory: false
  pair_type: judgement
  pairs:
    pop_low_mid:
      pos_desc: describe unpopular movies
      neg_desc: describe average movies
      pos_class:
      - pop_low
      neg_class:
      - pop_mid
      example_hypotheses:
      - is a documentary
      - is a remake
      - has a strange plot
    pop_high_mid:
      pos_desc: describe popular movies
      neg_desc: describe average movies
      pos_class:
      - pop_high
      neg_class:
      - pop_mid
      example_hypotheses:
      - mentions a superhero
      - talks more about surprise
      - is a horror plot
    revenue_low_mid:
      pos_desc: describe low-grossing movies
      neg_desc: describe average-grossing movies
      pos_class:
      - revenue_low
      neg_class:
      - revenue_mid
      example_hypotheses:
      - is a documentary
      - is a remake
      - has a strange plot
    revenue_high_mid:
      pos_desc: describe high-grossing movies
      neg_desc: describe average-grossing movies
      pos_class:
      - revenue_high
      neg_class:
      - revenue_mid
      example_hypotheses:
      - mentions a superhero
      - talks more about surprise
      - is a horror plot
  flip: false
- generation: how long the movie is
  applications:
  - target: how the length of the movie constrains the plot
    user: a film theorist
    example_hypotheses:
    - contains a plot twist with a betrayal
    - features a romance between the main characters
    - involves a heroic struggle against a powerful enemy
    v2-origid: 2184
    purely_exploratory: false
  pair_type: judgement
  pairs:
    runtime_low_mid:
      pos_desc: describe short movies
      neg_desc: describe average-length movies
      pos_class:
      - pop_low
      neg_class:
      - pop_mid
    runtime_high_mid:
      pos_desc: describe long movies
      neg_desc: describe average-length movies
      pos_class:
      - pop_high
      neg_class:
      - pop_mid
  flip: false
- generation: how big the movie's budget was
  applications:
  - target: the influence of funding on movie plot
    user: a film theorist
    example_hypotheses:
    - contains a plot twist with a betrayal
    - features a romance between the main characters
    - involves a heroic struggle against a powerful enemy
    v2-origid: 3184
    purely_exploratory: false
  pair_type: judgement
  pairs:
    runtime_low_mid:
      pos_desc: describe low-budget movies
      neg_desc: describe average-budget movies
      pos_class:
      - budget_low
      neg_class:
      - budget_mid
    runtime_high_mid:
      pos_desc: describe high-budget movies
      neg_desc: describe average-budget movies
      pos_class:
      - budget_high
      neg_class:
      - budget_mid
  flip: false
- generation: when the movie was released
  applications:
  - target: how movie plots have evolved over time
    user: a historian
    example_hypotheses:
    - describes a dystopian future with technology
    - features a romance between the main characters
    - involves a heroic struggle against a powerful enemy
    v2-origid: 4184
    purely_exploratory: false
  pair_type: judgement
  pairs:
    runtime_low_mid:
      pos_desc: describe movies from before 2000
      neg_desc: describe movies from after 2000
      pos_class:
      - pre_2000
      neg_class:
      - post_2000
  flip: FALSE
- generation: who produced the movie
  applications:
  - target: what types of movies are produced by our competitors
    user: a business analyst at a movie company
    example_hypotheses:
    - describes a dystopian future with technology
    - features a romance between the main characters
    - involves a heroic struggle against a powerful enemy
    v2-origid: 5184
    purely_exploratory: false
  pair_type: judgement
  pairs:
    company_mgm_warner:
      pos_desc: describe movies produced by MGM
      neg_desc: describe movies produced by Warner
      pos_class:
      - company_mgm
      neg_class:
      - company_warner
    company_paramount_fox:
      pos_desc: describe movies produced by Paramount
      neg_desc: describe movies produced by Fox
      pos_class:
      - company_paramount
      neg_class:
      - company_fox
    company_fox_universal:
      pos_desc: describe movies produced by Fox
      neg_desc: describe movies produced by Universal
      pos_class:
      - company_fox
      neg_class:
      - company_universal
    company_single_multiple:
      pos_desc: describe movies produced by multiple companies
      neg_desc: describe movies produced by a single company
      pos_class:
      - multiple_companies
      neg_class:
      - single_company
  flip: true
news_popularity:
- generation: popularity, which is measured by how many times the story was shared
  applications:
  - target: what specific topics cause people to read or share an article, so I know
      what to write about
    user: a journalist
    example_hypotheses:
    - mentions economic indicators, such as GDP and inflation
    - references President Trump's policies
    - mentions the potential risks of oil prices
    v2-origid: 188
    purely_exploratory: false
  pair_type: judgement
  pairs:
    economy_pop_unpop:
      pos_desc: are popular articles about the economy
      neg_desc: are unpopular articles about the economy
      pos_class:
      - economy_pop
      neg_class:
      - economy_unpop
      example_hypotheses:
      - is about an economic shock
      - talks about the U.S. economy
      - mentions gas prices
    microsoft_pop_unpop:
      pos_desc: are popular articles about Microsoft
      neg_desc: are unpopular articles about Microsoft
      pos_class:
      - microsoft_pop
      neg_class:
      - microsoft_unpop
      example_hypotheses:
      - talks about user privacy
      - talks about Zuckerberg
      - is disparaging
    obama_pop_unpop:
      pos_desc: are popular articles about Obama
      neg_desc: are unpopular articles about Obama
      pos_class:
      - obama_pop
      neg_class:
      - obama_unpop
      example_hypotheses:
      - talks about Obama negatively
      - talks about Afghanistan
      - mentions Obamacare
  flip: true
- generation: the positivity or negativity, measured by whether it received more likes
    or dislikes
  applications:
  - target: what specific topics are perceived positively, so I know what to write
      about
    user: a journalist
    example_hypotheses:
    - refers to Obama's health care policies, such as the Affordable Care Act
    - mentions Obama's efforts to address climate change
    - mentions Obama's foreign policy approach
    v2-origid: 191
    purely_exploratory: false
  pair_type: judgement
  pairs:
    obama_pos_neg:
      pos_desc: are positive articles about Obama
      neg_desc: are negative articles about Obama
      pos_class:
      - obama_pos
      neg_class:
      - obama_neg
      example_hypotheses:
      - talks about Obamacare
      - talks about the economy
  flip: true
nli_benchmarks:
- generation: which dataset the example is from
  applications:
  - target: any differences between these datasets which might influence performance
    user: a natural language processing researcher measuring language models on different
      tasks
    purely_exploratory: true
  pair_type: dataset
  pairs:
    anli_r1_r2_premise:
      pos_desc: are ANLI dataset Round 1 premises
      neg_desc: are  ANLI dataset Round 2 premises
      pos_class:
      - anli_r1_premise
      neg_class:
      - anli_r2_premise
      example_hypotheses: []
    mnli_match:
      pos_desc: are MNLI dataset hypotheses with mismatched annotations
      neg_desc: are MNLI dataset hypotheses with matched annotations
      pos_class:
      - mnli_mismatched_premise
      neg_class:
      - mnli_matched_premise
      example_hypotheses: []
    wanli_anil_r2_r3_premise:
      pos_desc: are ANLI dataset Round 2 premises
      neg_desc: are  ANLI dataset Round 3 premises
      pos_class:
      - anli_r2_premise
      neg_class:
      - anli_r3_premise
      example_hypotheses: []
    wanli_anli_premise:
      pos_desc: are WANLI dataset premises
      neg_desc: are ANLI dataset premises
      pos_class:
      - wanli_premise
      neg_class:
      - anli_r1_premise
      - anli_r2_premise
      - anli_r3_premise
      example_hypotheses: []
    enchmarks_wanli_anli_r2_r3_hypothesis:
      pos_desc: are ANLI dataset Round 1 hypotheses
      neg_desc: are  ANLI dataset Round 2 hypotheses
      pos_class:
      - anli_r1_hypothesis
      neg_class:
      - anli_r2_hypothesis
      example_hypotheses: []
    wanli_hans_premise:
      pos_desc: are WANLI dataset premises
      neg_desc: are HANS dataset premises
      pos_class:
      - wanli_premise
      neg_class:
      - hans_premise
      example_hypotheses: []
    wanli_mnli_premise:
      pos_desc: are WANLI dataset premises
      neg_desc: are MNLI dataset premises
      pos_class:
      - wanli_premise
      neg_class:
      - mnli_premise
      example_hypotheses: []
    wanli_qnli_hypothesis:
      pos_desc: are MNLI dataset hypotheses with mismatched annotations
      neg_desc: are MNLI dataset hypotheses with matched annotations
      pos_class:
      - mnli_mismatched_hypothesis
      neg_class:
      - mnli_matched_hypothesis
      example_hypotheses: []
    wanli_qnli_premise:
      pos_desc: are WANLI dataset premises
      neg_desc: are QNLI dataset premises
      pos_class:
      - wanli_premise
      neg_class:
      - qnli_premise
      example_hypotheses: []
  flip: true
npt_conferences:
- generation: which year the report was published
  applications:
  - target: what specific topics dominated NPT discussions across time
    user: a political scientist studying the history of the NPT
    example_hypotheses:
    - discusses the need for arms reduction talks between the US and Russia
    - calls for the accession of all states to the International Convention for the
      Suppression of Acts of Nuclear Terrorism and to the amended Convention on the
      Physical Protection of Nuclear Material
    - highlights the importance of the IAEA Comprehensive Safeguards as the NPT's
      only verification mechanism
    v2-origid: 202
    purely_exploratory: false
  pair_type: time
  pairs:
    pre_post_2008:
      pos_desc: are NPT conference reports before 2008
      neg_desc: are NPT conference reports between 2008 and 2012
      pos_class:
      - pre_2008
      neg_class:
      - btw_2008_2012
      example_hypotheses:
      - mentions Iran
      - mentions 9/11
      - talks less about the economy
    pre_post_2012:
      pos_desc: are NPT conference reports between 2008 and 2012
      neg_desc: are NPT conference reports after 2012
      pos_class:
      - btw_2008_2012
      neg_class:
      - post_2012
      example_hypotheses:
      - mentions Russia
      - is worried
      - talks about hypersonic missiles
  flip: true
open_deception:
- generation: whether subjects were told to lie
  applications:
  - target: the specific speaking styles of liars
    user: a rhetoric researcher studying the effect of lying on speaking style
    example_hypotheses:
    - uses descriptive language
    - uses language that is more specific than general
    - uses language that is more concrete than abstract
    v2-origid: 203
    purely_exploratory: false
  pair_type: intent
  pairs:
    lie_truth:
      pos_desc: are random truth statements
      neg_desc: are random false statements
      pos_class:
      - lie
      neg_class:
      - truth
      example_hypotheses:
      - talks about themselves
      - uses extreme adjectives
      - is a broad generalization
  flip: true
open_review:
- generation: the rating from paper reviewers
  applications:
  - target: the specific writing styles of good abstracts, so I know how to write
      my abstract
    user: a young researcher who hopes to get published
    example_hypotheses:
    - uses technical language, such as mathematical equations or scientific terms
    - concludes with a summary of the main findings
    - contains detailed, technical descriptions of a method or process
    v2-origid: 206
    purely_exploratory: false
  - target: the research areas of successful papers, so I know what to study
    user: a young researcher picking a topic
    example_hypotheses:
    - proposes a method for learning interpretable representations
    - uses language to transfer natural language methods
    - utilizes multi-agent reinforcement learning techniques
    v2-origid: 207
    purely_exploratory: false
  pair_type: judgement
  pairs:
    good_bad:
      pos_desc: are good journal submissions
      neg_desc: are bad journal submissions
      pos_class:
      - good_papers
      neg_class:
      - bad_papers
      example_hypotheses:
      - is clearer
      - provides more equations
      - talks about existing literature
    great_good:
      pos_desc: are good journal submissions
      neg_desc: are very good journal submissions
      pos_class:
      - great_papers
      neg_class:
      - good_papers
      example_hypotheses:
      - expresses surprise
      - introduces a benchmark
      - introduces a new technique
  flip: false
oral_histories:
- generation: the race of the narrator
  applications:
  - target: the specific struggles of people of different races
    user: a sociologist studying how race affects living conditions
    example_hypotheses:
    - describes stories of sexual harassment/assault in the minority community
    - mentions the State's role in violence
    - mentions the prevalence of child sexual abuse
    v2-origid: 208
    purely_exploratory: false
  - target: how people of different races feel about America
    user: a sociologist studying how race affects living conditions
    example_hypotheses:
    - expresses spite and hatred towards the government
    - discusses the importance of self-advocacy and resilience to survive
    - is optimistic about gaining rights
    v2-origid: 2208
    purely_exploratory: false
  pair_type: demographic
  pairs:
    asian_white:
      pos_desc: are oral histories of Asian people
      neg_desc: are oral histories of white people
      pos_class:
      - asian
      neg_class:
      - white
      example_hypotheses:
      - talks more about racism
      - talks more about civil rights
      - mentions jobs
    black_white:
      pos_desc: are oral histories of black people
      neg_desc: are oral histories of white people
      pos_class:
      - black
      neg_class:
      - white
      example_hypotheses:
      - talks more about racism
      - talks more about civil rights
      - discusses violence
  flip: true
- generation: the education level of the narrator
  applications:
  - target: the specific experiences and struggles of people of different education
      levels
    user: a sociologist studying how education affects living conditions
    example_hypotheses:
    - mentions more experiences of urban or metropolitan settings
    - mentions professional experiences such as jobs, internships, and volunteer work
    - talks about travel experiences
    v2-origid: 210
    purely_exploratory: false
  pair_type: demographic
  pairs:
    college_educated:
      pos_desc: are oral histories of people with college degrees
      neg_desc: are oral histories of people without college degrees
      pos_class:
      - college_educated
      neg_class:
      - not_college_educated
      example_hypotheses:
      - talks more about academic jobs
      - mentions civil rights
      - uses eloquent language
      - talks about protest
  flip: true
- generation: when the narrator lived
  applications:
  - target: the specific experiences and struggles of people across time
    user: a historian writing about everyday people in different time periods
    example_hypotheses:
    - mentions an instance of touching someone without consent
    - talks about the civil rights movement and its progress
    - mentions the difficulties of finding or keeping a job
    v2-origid: 211
    purely_exploratory: false
  pair_type: time
  pairs:
    pre_post_1930:
      pos_desc: are oral histories of people born before 1930
      neg_desc: are oral histories of people born between 1930 and 1950
      pos_class:
      - pre_1930
      neg_class:
      - 1930-50
      example_hypotheses:
      - talks more about racism
      - talks about older presidents
      - mentions World War I
    pre_post_1950:
      pos_desc: are oral histories of people born between 1930 and 1950
      neg_desc: are oral histories of people born after 1950
      pos_class:
      - 1930-50
      neg_class:
      - post_1950
      example_hypotheses:
      - talks about the Great Depression
      - mentions World War II
      - talks about jobs
  flip: true
- generation: where the narrator lived
  applications:
  - target: the specific experiences and struggles of people in different regions
      of the United States
    user: a historian studying life in different regions of the United States
    example_hypotheses:
    - mentions the importance of family and traditional values
    - refers to the Ku Klux Klan and other hate groups
    - mentions the influence of religion and culture
    v2-origid: 213
    purely_exploratory: false
  pair_type: location
  pairs:
    south_not_south:
      pos_desc: are oral histories of people from the South
      neg_desc: are oral histories of people not from the South
      pos_class:
      - south
      neg_class:
      - not_south
      example_hypotheses:
      - talks more about discrimination
      - talks about economic struggles
      - mentions migration
  flip: true
parenting_reddit_users:
- generation: what year the post was made
  applications:
  - target: how parents' concerns and needs have changed over the years
    user: a sociologist studying parenting
    example_hypotheses:
    - mentions the use of technology, such as video calls and online resources
    - discusses the difficulty of finding childcare options due to COVID-19
    - discusses the importance of wifi safety
    v2-origid: 217
    purely_exploratory: false
  pair_type: demographic
  pairs:
    year_2016_2017:
      pos_desc: are Reddit posts from 2017
      neg_desc: are Reddit posts from 2016
      pos_class:
      - year_2017
      neg_class:
      - year_2016
      example_hypotheses:
      - argues for importance of breastfeeding
    year_2017_2018:
      pos_desc: are Reddit posts from 2018
      neg_desc: are Reddit posts from 2017
      pos_class:
      - year_2018
      neg_class:
      - year_2017
      example_hypotheses:
      - argues for importance of breastfeeding
    year_2018_2019:
      pos_desc: are Reddit posts from 2019
      neg_desc: are Reddit posts from 2018
      pos_class:
      - year_2019
      neg_class:
      - year_2018
      example_hypotheses:
      - argues for importance of breastfeeding
    year_2019_2020:
      pos_desc: are Reddit posts from 2020
      neg_desc: are Reddit posts from 2019
      pos_class:
      - year_2020
      neg_class:
      - year_2019
      example_hypotheses:
      - argues for importance of breastfeeding
  flip: true
- generation: whether the parent is a mother or father
  applications:
  - target: the specific concerns and needs of mothers versus fathers
    user: a sociologist studying parenting
    example_hypotheses:
    - expresses anxiety about leaving children with a caregiver
    - discusses the need for preventive measures to protect children's health
    - mentions worries about not having enough energy to take care of children
    v2-origid: 218
    purely_exploratory: false
  - target: which topics mothers and fathers each bring up
    user: a sociologist studying parenting
    example_hypotheses:
    - mentions the child's education and intellectual growth
    - talks about potty training
    - brings up celebrations, such as a gender reveal
    v2-origid: 218
    purely_exploratory: false
  pair_type: demographic
  pairs:
    moms_dads:
      pos_desc: are Reddit posts from mothers
      neg_desc: are Reddit posts from fathers
      pos_class:
      - moms
      neg_class:
      - dads
      example_hypotheses:
      - talks about nursing
      - talks about breastfeeding
      - mentions problems with sleep
  flip: true
- generation: how long the user has been posting on parenting subreddits
  applications:
  - target: how different concerns and needs come up as a baby grows
    user: a sociologist studying the process of raising a child
    example_hypotheses:
    - talks about balancing work and family life
    - mentions the use of CIO (Cry It Out) methods for newborns
    - mentions the use of verbal calming techniques to defuse fights
    v2-origid: 221
    purely_exploratory: false
  pair_type: demographic
  pairs:
    month_1_month_3:
      pos_desc: are posts from parents who have beem on Reddit for less than a month
      neg_desc: are parenting posts from the first to third months on Reddit
      pos_class:
      - new
      - 30_days
      neg_class:
      - 90_days
      example_hypotheses:
      - talks about giving birth
      - mentions a hospital
    month_3_year_1:
      pos_desc: are Reddit posts from parents first 3 months on Reddit
      neg_desc: are parenting Reddit posts from 3 months to 1 year
      pos_class:
      - new
      - 30_days
      - 90_days
      neg_class:
      - half_year
      - 1_year
      example_hypotheses:
      - talks more about breastfeeding
      - mentions a newborn
    year_2_year_4:
      pos_desc: are Reddit posts on parenting subreddits from accounts more than 5
        years old
      neg_desc: are Reddit posts on parenting subreddits from accounts less than 5
        years old
      pos_class:
      - new
      - 30_days
      - 90_days
      - half_year
      - 1_year
      - 2_years
      - 3_years
      - 4_years
      neg_class:
      - 5_years
      example_hypotheses:
      - gives advice
      - talks about school
  flip: true
parenting_subreddits:
- generation: which Subreddit the post is from
  applications:
  - target: the specific concerns and needs that come up on different topics
    user: a parent trying to figure out which forum to post on
    example_hypotheses:
    - discusses mental health issues such as depression and anxiety
    - mentions struggles with relationships with co-parent
    - discusses loneliness and lack of social support
    v2-origid: 223
    purely_exploratory: false
  pair_type: subject
  pairs:
    breastfeeding_baby_food:
      pos_desc: are Reddit posts froma Subreddit about breastfeeding
      neg_desc: are Reddit posts from a Subreddit about baby food
      pos_class:
      - breastfeeding
      neg_class:
      - baby food
      example_hypotheses:
      - mentions feeling tired
      - talks about younger children
    parent_support_interaction:
      pos_desc: are Reddit posts from a Subreddit about parents asking for support
      neg_desc: are Reddit posts from a Subreddit about interacting with children
      pos_class:
      - parent support
      neg_class:
      - parent-child interaction
      example_hypotheses:
      - is concerned about work
      - mentinos mental health
    single_non_biological_parents:
      pos_desc: are Reddit posts from a Subreddit about single parents
      neg_desc: are Reddit posts from a Subreddit about non-biological parents
      pos_class:
      - sinle parents
      neg_class:
      - non-biological parents
      example_hypotheses:
      - talks more about finances
      - mentions missing a father
  flip: true
poetry:
- generation: who wrote the poem
  applications:
  - target: the specific topics that each poet writes about
    user: a comparative literature researcher
    example_hypotheses:
    - uses personification to give life to inanimate objects and animals
    - depicts a longing for the past
    - contains references to mythology and classical literature
    v2-origid: 226
    purely_exploratory: false
  pair_type: authorship
  pairs:
    poet_dickinson_tennyson:
      pos_desc: are poems by Emily Dickinson
      neg_desc: are poems by Alfred, Lord Tennyson
      pos_class:
      - poet_dickinson
      neg_class:
      - poet_tennyson
      example_hypotheses:
      - discusses death
      - talks about identity
    poet_wordsworth_shakespeare:
      pos_desc: are poems by William Wordsworth
      neg_desc: are poems by William Shakespeare
      pos_class:
      - poet_wordsworth
      neg_class:
      - poet_shakespeare
      example_hypotheses:
      - talks about nature
      - comments on the nature of human experience
  flip: true
- generation: the subject of the poem
  applications:
  - target: the specific tones, themes, and motifs associated with each subject
    user: a comparative literature researcher
    example_hypotheses:
    - discusses the consequences of sin
    - includes descriptions of nature or the outdoors
    - explores the idea of hope and possibility
    v2-origid: 228
    purely_exploratory: false
  pair_type: subject
  pairs:
    subject_commentary_history:
      pos_desc: are poems that contain social commentary
      neg_desc: are poems about history
      pos_class:
      - subject_commentary
      neg_class:
      - subject_history
      example_hypotheses:
      - is more opinionated
      - is judgemental
      - mentions injustice
    subject_living_death:
      pos_desc: are poems about living
      neg_desc: are poems about death
      pos_class:
      - subject_living
      neg_class:
      - subject_death
      example_hypotheses:
      - are more positive in tone
      - talks about joy
    subject_relationships_family:
      pos_desc: are poems about relationships
      neg_desc: are poems about family
      pos_class:
      - subject_relationships
      neg_class:
      - subject_family
      example_hypotheses:
      - alludes to sex
      - does not mention child
    subject_time_nature:
      pos_desc: are poems about time and brevity
      neg_desc: are poems about nature
      pos_class:
      - subject_time
      neg_class:
      - subject_nature
      example_hypotheses:
      - discuss humans
      - is regretful
  flip: true
political_ads:
- generation: the age range of the ad's target demographic
  applications:
  - target: what messages political organizations send to children
    user: a political scientist studying media
    example_hypotheses:
    - talks about humanitarain causes
    - mentions the need for families
    - advocates for political unity and bipartisanship
    v2-origid: 2000
    purely_exploratory: false
  pair_type: demographic
  flip: true
  pairs:
    adult_nonadults:
      pos_desc: are shown to adults, teenagers, and children
      neg_desc: are shown to only adults
      pos_class:
      - age_children
      neg_class:
      - age_adults
    teenagers_children:
      pos_desc: are shown to adults, teenagers, and children
      neg_desc: are shown to only adults and teenagers
      pos_class:
      - age_teenagers
      neg_class:
      - age_children
- generation: the gender of the ad's target demographic
  applications:
  - target: how political organizations appeal to men versus women
    user: a political scientist studying gender and politics
    example_hypotheses:
    - talks about humanitarain causes
    - mentions the need for families
    - pushes for a stronger military
    - advocates for political unity and bipartisanship
    v2-origid: 2001
    purely_exploratory: false
  pair_type: demographic
  flip: true
  pairs:
    female_any:
      pos_desc: are shown to only females
      neg_desc: are shown to any genders
      pos_class:
      - gender_female
      neg_class:
      - gender_any
    male_any:
      pos_desc: are shown to only males
      neg_desc: are shown to any genders
      pos_class:
      - gender_male
      neg_class:
      - gender_any
- generation: the country the ad is targeted at
  applications:
  - target: differences in political advertisement strategy across countries
    user: a political scientist studying comparative government
    example_hypotheses:
    - asks for a small donation
    - emphasizes the threat to the country's safety
    - demonizes the political opponent
    v2-origid: 2003
    purely_exploratory: false
  - target: policy priorities of parties in each country
    user: a political scientist studying comparative government
    example_hypotheses:
    - mentions the threat of global warming
    - pushes back against efforts to contain COVID-19
    - brings up terrorism and other military threats
    v2-origid: 2004
    purely_exploratory: false
  pair_type: location
  flip: true
  pairs:
    us_cad:
      pos_desc: are shown to Americans
      neg_desc: are shown to Canadians
      pos_class:
      - location_us
      neg_class:
      - location_cad
    uk_aus:
      pos_desc: are shown to UK citizens
      neg_desc: are shown to Australians
      pos_class:
      - location_uk
      neg_class:
      - location_aus
    uk_us:
      pos_desc: are shown to UK citizens
      neg_desc: are shown to Americans
      pos_class:
      - location_uk
      neg_class:
      - location_us
- generation: the state the ad is targeted at
  applications:
  - target: how parties target voters in different states
    user: an American political scientist studying election messaging
    example_hypotheses:
    - brings up past mistakes of the opponent
    - mentions issues related to abortion
    - appeals to threats to democracy itself
    v2-origid: 2005
    purely_exploratory: false
  pair_type: demographic
  flip: true
  pairs:
    ca_ny:
      pos_desc: are shown to Californians
      neg_desc: are shown to New Yorkers
      pos_class:
      - location_ca
      neg_class:
      - location_nyc
    ga_tx:
      pos_desc: are shown to Georgians
      neg_desc: are shown to Texans
      pos_class:
      - location_ga
      neg_class:
      - location_tx
    ca_ga:
      pos_desc: are shown to Georgians
      neg_desc: are shown to Californians
      pos_class:
      - location_ga
      neg_class:
      - location_ca
- generation: the reported interests of the ad's target demographic
  applications:
  - target: the arguments each party makes for each interest area
    user: a political journalists reporting on party strategy
    example_hypotheses:
    - appeals to abortion rights
    - disparages critical race theory in schools
    - mentions a rise in crime under current leadership
    v2-origid: 2006
    purely_exploratory: false
  pair_type: demographic
  flip: true
  pairs:
    social_issues:
      pos_desc: are ads targeting social issues from the Democratic party
      neg_desc: are ads targeting social issues from the Republican party
      pos_class:
      - interest_social_issues_dem
      neg_class:
      - interest_social_issues_rep
    politics:
      pos_desc: are ads targeting politics from the Democratic party
      neg_desc: are ads targeting politics issues from the Republican party
      pos_class:
      - interest_politics_dem
      neg_class:
      - interest_politics_rep
    community_issues:
      pos_desc: are ads targeting community issues from the Democratic party
      neg_desc: are ads targeting community issues issues from the Republican party
      pos_class:
      - interest_community_issues_dem
      neg_class:
      - interest_community_issues_rep
    education:
      pos_desc: are ads targeting education from the Democratic party
      neg_desc: are ads targeting education issues from the Republican party
      pos_class:
      - interest_education_dem
      neg_class:
      - interest_education_rep
    family:
      pos_desc: are ads targeting family from the Democratic party
      neg_desc: are ads targeting family issues from the Republican party
      pos_class:
      - interest_family_dem
      neg_class:
      - interest_family_rep
- generation: when the ad was shown
  applications:
  - target: which messages were salient in the U.S. during each period
    user: an American journalist tracking political messaging over time
    example_hypotheses:
    - appeals to abortion rights
    - disparages critical race theory in schools
    - mentions a rise in crime under current leadership
    v2-origid: 2007
    purely_exploratory: false
  pair_type: time
  flip: true
  pairs:
    janmar_aprjun:
      pos_desc: are American ads shown between January and March 2021
      neg_desc: are American ads shown between July and September 2021
      pos_class:
      - time_21_jan_mar
      neg_class:
      - time_21_jul_sep
radiology_diagnosis:
- generation: whether medical experts agreed on a diagnosis
  applications:
  - target: what symptoms make experts more confident in a diagnosis
    user: a medical student training to become a radiologist
    example_hypotheses:
    - describe mild or scattered lung densities
    - clearly describes the lack of pneumonic abnormality
    - mentions pleural effusion or adenopathy
    v2-origid: 232
    purely_exploratory: false
  pair_type: judgement
  pairs:
    all_consensus_impression:
      pos_desc: describe impressions of patients with consensus diagnoses
      neg_desc: describe impressions of patients with conflicting diagnoses
      pos_class:
      - all_consensus_impression
      neg_class:
      - all_no_consensus_impression
      example_hypotheses:
      - mentions a fever
      - lists many symptoms
    cough_consensus_impression:
      pos_desc: describe impressions of cough patients with consensus diagnoses
      neg_desc: describe impressions of cough patients with conflicting diagnoses
      pos_class:
      - cough_consensus_impression
      neg_class:
      - cough_no_consensus_impression
      example_hypotheses:
      - dismisses more severe diseases
      - does not mention fever
  flip: true
- generation: which diagnosis the patient was assigned
  applications:
  - target: which symptoms or histories suggest each diagnosis
    user: a medical student training to become a radiologist
    example_hypotheses:
    - mentions a history of walking pneumonia
    - mentions a repair of a duodenal web within the last month
    - mentions a cough and congestion for a week
    v2-origid: 233
    purely_exploratory: false
  pair_type: judgement
  pairs:
    cough_fever_history:
      pos_desc: describe medical histories of patients with just a cough
      neg_desc: describe medical histories of patients with a fever
      pos_class:
      - cough_history
      neg_class:
      - fever_history
      example_hypotheses:
      - mentions fewer comorbidities
      - describes young patient
    cough_pneumonia_impression:
      pos_desc: describe impressions of patients with a cough
      neg_desc: describe impressions of patients with a pneumonia, a lung inflamation
      pos_class:
      - cough_impression
      neg_class:
      - pneumonia_impression
      example_hypotheses:
      - suggests less severe cough
      - does not mention fever
    uti_vesicoureteral_reflux_impression:
      pos_desc: describe impressions of patients with a urinary track infection
      neg_desc: describe impressions of patients with vesicoureteral reflux, an abnormal
        flow of urine
      pos_class:
      - uti_impression
      neg_class:
      - vesicoureteral_reflux_impression
      example_hypotheses:
      - discusses sources of infection
      - does not mention fever
  flip: true
rate_my_prof:
- generation: whether the lecturer is male or female
  applications:
  - target: how a lecturer's gender affects what people bring up in reviews
    user: a university dean worried about gender bias on review sites
    example_hypotheses:
    - mentions the lecturer's physical attractiveness
    - mentions the lecturer's use of humor
    - mentions the lecturer's grading policy
    v2-origid: 236
    purely_exploratory: false
  pair_type: demographic
  pairs:
    female_male:
      pos_desc: rate female lecturers
      neg_desc: rate male lecturers
      pos_class:
      - female
      neg_class:
      - male
      example_hypotheses:
      - expresses more disatisfaction
      - comments on lecturing style
      - complains of being annoyed
      - claims that professor is unqualified
  flip: true
reddit_humor:
- generation: whether the joke recieved a lot of upvotes
  applications:
  - target: the specific topics and setups that people find funny, so I can write
      a funny joke
    user: an aspiring comedian
    example_hypotheses:
    - pokes fun at an everyday situation
    - touches on a taboo subject in a lighthearted manner
    - uses self-deprecating humor
    v2-origid: 237
    purely_exploratory: false
  pair_type: judgement
  pairs:
    funny_unfunny:
      pos_desc: are funny Reddit jokes
      neg_desc: are unfunny Reddit jokes
      pos_class:
      - funny
      neg_class:
      - unfunny
      example_hypotheses:
      - has a long buildup
      - has a surprising punchline
      - is not about race
  flip: true
reddit_stress:
- generation: which Subreddit the post was submitted to
  applications:
  - target: the cause of different kinds of stress
    user: training to become a psychiatrist
    example_hypotheses:
    - mentions financial problems, such as lack of money or difficulty affording medication
    - mentions a fear of fainting or other physical reactions
    - mentions difficulty in the college application process, such as not having enough
      time for essays, not finishing college lists, or not having enough money for
      tuition
    v2-origid: 238
    purely_exploratory: false
  pair_type: subject
  pairs:
    anxiety_stress:
      pos_desc: are posts from a subreddit about anxiety
      neg_desc: are posts from a subreddit about stress
      pos_class:
      - anxiety
      neg_class:
      - stress
      example_hypotheses:
      - talks about the future
      - mentions social standing
      - talks more about depression
    ptsd_anxiety:
      pos_desc: are posts from a subreddit about PTSD
      neg_desc: are posts from a subreddit about anxiety
      pos_class:
      - ptsd
      neg_class:
      - anxiety
      example_hypotheses:
      - mentions military service
      - mentions trauma
      - talks about noises
    abuse_social:
      pos_desc: are posts from subreddits about abuse
      neg_desc: are posts from subreddits about relationship problems
      pos_class:
      - abuse
      neg_class:
      - social
    stress_financial:
      pos_desc: are posts from subreddits about stress and anxiety generally
      neg_desc: are posts from subreddits about financial problems
      pos_class:
      - anxiety
      - stress
      neg_class:
      - financial
  flip: true
reuters_authorship:
- generation: which Reuters journalist wrote the article
  applications:
  - target: the topics that each journalist covers
    user: trying to figure out who wrote this anonymous news article
    example_hypotheses:
    - mentions the legal aspect of a situation, such as court cases or regulatory
      issues
    - discusses industry-wide changes, such as consolidation or joint ventures
    - features quotes from CEOs or other figures in the industry
    v2-origid: 244
    purely_exploratory: false
  - target: the specific writing style of each journalist
    user: trying to figure out who wrote this anonymous news article
    example_hypotheses:
    - uses a formal and professional tone
    - uses specific financial figures, such as '$5 billion'
    - uses the phrase 'raising the stakes.'
    v2-origid: 245
    purely_exploratory: false
  pair_type: authorship
  pairs:
    janlopatka_john_mastrini:
      pos_desc: are news articles by Jan Lopatka
      neg_desc: are news articles by John Mastrini
      pos_class:
      - JanLopatka
      neg_class:
      - JohnMastrini
      example_hypotheses:
      - talks more about politics
      - talks less about the economy
      - expresses more political views
    lynneodonnel_sarahdavison:
      pos_desc: are news articles by Lynne O'Donnel
      neg_desc: are news articles by Sarah Davidson
      pos_class:
      - LynneO'Donnell
      neg_class:
      - SarahDavison
      example_hypotheses:
      - talks more about the economy
      - uses fewer quotes
      - expresses greater uncertainty
    robinsidel_bernardhickey:
      pos_desc: are news articles by Robin Sidel
      neg_desc: are news articles by Bernard Hickey
      pos_class:
      - RobinSidel
      neg_class:
      - BernardHickey
      example_hypotheses:
      - uses more quotes
      - contains longer sentences
      - talks about the stock market
  flip: true
riddles:
- generation: whether the words fit a secret rule
  applications:
  - target: something about the spelling of the words that explains the difference
    user: trying to solve this riddle with my friends
    example_hypotheses:
    - has at least three syllables
    - contains a letter that occurs more than once in the word
    - contains words with a 'p' as the fourth letter
    v2-origid: 247
    purely_exploratory: false
  pair_type: interpretability
  pairs:
    ggd:
      pos_desc: are words that pass
      neg_desc: are words that don't pass
      pos_class:
      - ggd_pos
      neg_class:
      - ggd_neg
      example_hypotheses:
      - rhymes with orange
      - is a palindrome
    t:
      pos_desc: are words that pass
      neg_desc: are words that don't pass
      pos_class:
      - t_pos
      neg_class:
      - t_neg
      example_hypotheses:
      - doesn't contains the letter S
      - starts with P
  flip: true
scotus_cases:
- generation: the relationship between the SCOTUS decision and the decision of the
    lower court
  applications:
  - target: which policy topics cause the SCOTUS to disagree with lower courts
    user: a political scientist studying Supreme Court rulings
    example_hypotheses:
    - involve cases in which the plaintiff is from a minority group
    - involve issues of federalism, such as the division of powers between state and
      federal governments
    - involve the interpretation of international treaties or agreements
    v2-origid: 249
    purely_exploratory: false
  pair_type: judgement
  pairs:
    affirm_reverse:
      pos_desc: are facts from Supreme Court cases where a lower court's ruling was
        reversed
      neg_desc: are facts from Supreme Court cases where a lower court's ruling was
        affirmed
      pos_class:
      - dispo_reversed
      neg_class:
      - dispo_affirmed
      example_hypotheses:
      - is about jurisdiction
      - mentions civil rights
    vacated_reverse:
      pos_desc: are facts from Supreme Court cases where a lower court's ruling was
        vacated
      neg_desc: are facts from Supreme Court cases where a lower court's ruling was
        reversed
      pos_class:
      - dispo_vacated_remanded
      neg_class:
      - dispo_reversed
      example_hypotheses:
      - is about a civil suit
      - is less strong in wording
  flip: true
- generation: who was president when the case was heard
  applications:
  - target: which policy topics were brought up before the court during each presidency
    user: a political scientist studying Supreme Court rulings
    example_hypotheses:
    - involves tort claims, such as the Torture Victim Protection Act
    - concerns patent law or intellectual property
    - talks about civil rights issues, such as discrimination
    v2-origid: 250
    purely_exploratory: false
  pair_type: time
  pairs:
    obama_trump:
      pos_desc: are facts from Supreme Court cases during the Obama presidency
      neg_desc: are facts from Supreme Court cases during the Trump presidency
      pos_class:
      - era_pres_obama
      neg_class:
      - era_pres_trump
      example_hypotheses:
      - addresses issues about economic activity
      - mentions criminal disputes
      - is about LGBT issues
  flip: true
- generation: who was Chief Justice at the time of the case
  applications:
  - target: which policy areas were brought up before the court under each Chief Justice
    user: a political scientist studying Supreme Court rulings
    example_hypotheses:
    - involve civil rights, such as those of disabled individuals
    - involves First Amendment rights
    - involves the right to a fair trial
    v2-origid: 251
    purely_exploratory: false
  pair_type: time
  pairs:
    burger_rehnquist:
      pos_desc: are facts from cases heard by the Rehnquist court
      neg_desc: are facts from cases heard by the Roberts court
      pos_class:
      - era_chief_rehnquist
      neg_class:
      - era_chief_roberts
      example_hypotheses:
      - addresses traditionally conservative topics
      - mentions civil rights
    criminal_rehnquist_roberts:
      pos_desc: are facts from criminal Supreme Court cases heard by the Roberts court
      neg_desc: are facts from criminal Supreme Court cases heard by the Rehnquist
        court
      pos_class:
      - issue_criminal_era_chief_roberts
      neg_class:
      - issue_criminal_era_chief_rehnquist
      example_hypotheses:
      - mentions the death penalty
      - talks about due process
    economic_rehnquist_roberts:
      pos_desc: are facts from civil rights Supreme Court cases heard by the Warren
        court
      neg_desc: are facts from civil rights Supreme Court cases heard by the Burger
        court
      pos_class:
      - issue_civil_era_chief_warren
      neg_class:
      - issue_civil_era_chief_burger
      example_hypotheses:
      - talks about blatant discrimination
      - does not mention the Civil Rights Act
    warren_burger:
      pos_desc: are facts from cases heard by the Warren court
      neg_desc: are facts from cases heard by the Burger court
      pos_class:
      - era_chief_warren
      neg_class:
      - era_chief_burger
      example_hypotheses:
      - is about a civil rights violation
      - mentions racism
  flip: true
- generation: who won the case
  applications:
  - target: the types of complains that the first party is more likely to win
    user: a lawyer preparing a case in front of the Supreme Court
    example_hypotheses:
    - describe a voluntary action taken by the first party, such as a petition or
      a settlement agreement
    - talks about constitutional rights, such as the Fourth Amendment
    - has evidence of a deliberate attempt to violate a law or regulation
    v2-origid: 255
    purely_exploratory: false
  pair_type: judgement
  pairs:
    first_party_win:
      pos_desc: are facts from Supreme Court cases where the first party won
      neg_desc: are facts from Supreme Court cases where the first party lost
      pos_class:
      - first_party_win
      neg_class:
      - first_party_lose
      example_hypotheses:
      - describes a violation of civil rights
      - mentions state law
  flip: true
- generation: whether the justices agreed with one another
  applications:
  - target: which policy areas the justices can reach consensus on
    user: a political scientist studying Supreme Court rulings
    example_hypotheses:
    - involves a defendant who was accused of violating an obscenity law
    - involve a challenge under the Fourth Amendment
    - attempts to balance individual rights and the interests of the state
    v2-origid: 256
    purely_exploratory: false
  pair_type: judgement
  pairs:
    plurality:
      pos_desc: are facts from Supreme Court cases with only a plurality (no majority)
        decision
      neg_desc: are facts from Supreme Court cases with a majority decision
      pos_class:
      - decision_plurality
      neg_class:
      - decision_majority
      example_hypotheses:
      - describes a partisan topic
      - talks about guns
      - talks about religious liberties
    unanimous:
      pos_desc: are facts from Supreme Court cases with a unanimous decision
      neg_desc: are facts from Supreme Court cases with only a majority decision
      pos_class:
      - decision_per_curiam
      neg_class:
      - decision_majority
      example_hypotheses:
      - is not about a partisan issue
      - describes egregious rights violations
  flip: true
- generation: who the parties are
  applications:
  - target: which general policy areas come up
    user: a political scientist studying Supreme Court rulings
    example_hypotheses:
    - involves a dispute over the selection of a jury
    - involves a dispute over the protection of civil rights
    - involves the use of surveillance evidence
    v2-origid: 258
    purely_exploratory: false
  - target: which specific laws and regulations are litigated
    user: a political scientist studying Supreme Court rulings
    example_hypotheses:
    - discusses the application of the 14th Amendment
    - involves the application of the Equal Protection Clause
    - involve cases that involve the interpretation of the Low-Level Radioactive Waste
      Management Act Amendments
    v2-origid: 2258
    purely_exploratory: false
  pair_type: subject
  pairs:
    state_first_second:
      pos_desc: are facts from Supreme Court cases involving an American state as
        the plaintiff
      neg_desc: are facts from Supreme Court cases involving an American state as
        the defendant
      pos_class:
      - party_1st_state
      neg_class:
      - party_2nd_state
      example_hypotheses:
      - describes state crimes
      - is about inter-state conflict
    state_or_not:
      pos_desc: are facts from Supreme Court cases involving an American state as
        a party
      neg_desc: are facts from Supreme Court cases without an American state as a
        party
      pos_class:
      - party_1st_state
      - party_2nd_state
      neg_class:
      - party_neither_state
      example_hypotheses:
      - describes challenges to state policies
      - mentions a state law
    us_first_second:
      pos_desc: are facts from Supreme Court cases involving the United States as
        the plaintiff
      neg_desc: are facts from Supreme Court cases involving the United States as
        the defendant
      pos_class:
      - party_1st_us
      neg_class:
      - party_2nd_us
      example_hypotheses:
      - describes federal crimes
      - mentions murder
    us_or_not:
      pos_desc: are facts from Supreme Court cases involving the United States as
        a party
      neg_desc: are facts from Supreme Court cases without the United States as a
        party
      pos_class:
      - party_1st_us
      - party_2nd_us
      neg_class:
      - party_neither_us
      example_hypotheses:
      - describes nationally relevant topics
      - is about a felony
  flip: true
short_answer_scoring:
- generation: how good the score was
  applications:
  - target: the specific writing style of good writers, so that I can improve my own
      writing
    user: a student who has to write several responses toshort answer questions
    example_hypotheses:
    - makes use of active verbs and phrases
    - uses more precise vocabulary
    - uses facts and examples to back up statements
    v2-origid: 262
    purely_exploratory: false
  pair_type: judgement
  pairs:
    good_bad:
      pos_desc: are good short answers
      neg_desc: are bad short answers
      pos_class:
      - good_answers
      neg_class:
      - bad_answers
      example_hypotheses:
      - uses correct grammar
      - gives more detail
      - uses longer words
    good_medium:
      pos_desc: are good short answers
      neg_desc: are average short answers
      pos_class:
      - good_answers
      neg_class:
      - medium_answers
      example_hypotheses:
      - gives more examples
      - writes in complete sentences
      - has good grammar
  flip: false
stock_news:
- generation: the state of the stock market on the day of the headline
  applications:
  - target: what specific news events cause a stock market rally
    user: a quantitative trader trying to beat the market
    example_hypotheses:
    - highlights a major technological breakthrough
    - references a successful military operation
    - references celebrations or anniversaries
    v2-origid: 264
    purely_exploratory: false
  pair_type: socioeconomic
  pairs:
    up_down:
      pos_desc: are headlines on days the stock market rises
      neg_desc: are headlines on days the stock market falls
      pos_class:
      - up
      neg_class:
      - down
      example_hypotheses:
      - talks about a bill being passed
      - talks about economic growth
      - mention a war ending
  flip: true
suicide_notes:
- generation: the suicidal intent of the author
  applications:
  - target: the specific topics or tones brought up by people who actually plan to
      commit suicide
    user: a psychiatrist hoping to better help my patients
    example_hypotheses:
    - mentions taking medication, or going to therapy/counseling
    - mentions feeling ashamed or embarrassed
    - mentions a lack of support system
    v2-origid: 265
    purely_exploratory: false
  pair_type: intent
  pairs:
    suicide_depression:
      pos_desc: are notes from people who are suicidal
      neg_desc: are notes from people who are depressed, but not suicidal
      pos_class:
      - suicide
      neg_class:
      - depression
      example_hypotheses:
      - is more apologetic
      - mentions family and friends
      - is more thorough
      - uses better grammar
  flip: true
times_india_headlines:
- generation: the year the headline was written
  applications:
  - target: which topics dominated the news from year to year
    user: an Indian historian writing about recent trends
    example_hypotheses:
    - mentions the economic effects of deficient rain
    - mentions the possible formation of an alliance in Uttar Pradesh
    - mentions crimes, such as robberies and murders
    v2-origid: 268
    purely_exploratory: false
  pair_type: time
  pairs:
    '2003_2004':
      pos_desc: are Indian news headlines from 2003
      neg_desc: are Indian news headlines from 2004
      pos_class:
      - '2003'
      neg_class:
      - '2004'
      example_hypotheses:
      - talks more about iraq
      - mentions train crashes
      - talks about Nadimarg massacre
    '2007_2008':
      pos_desc: are Indian news headlines from 2007
      neg_desc: are Indian news headlines from 2008
      pos_class:
      - '2007'
      neg_class:
      - '2008'
      example_hypotheses:
      - talks less about the economy
      - is more optimistic
    '2019_2020':
      pos_desc: are Indian news headlines from 2019
      neg_desc: are Indian news headlines from 2020
      pos_class:
      - '2019'
      neg_class:
      - '2020'
      example_hypotheses:
      - does not mention COVID-19
      - is more optimistic
  flip: true
trial_deception:
- generation: whether the testimony turned out to be deceptive
  applications:
  - target: the specific claims of innocent people
    user: a judge concerned about false testimony
    example_hypotheses:
    - contains descriptions of events that are easy to understand and follow
    - mentions trying to de-escalate a situation
    - mentions being in a state of shock after the incident
    v2-origid: 269
    purely_exploratory: false
  - target: the general tone of innocent people
    user: a judge concerned about false testimony
    example_hypotheses:
    - includes statements of gratitude and appreciation for a particular person or
      thing
    - expresses appreciation for the support and kindness of others
    - contains language of regret or remorse
    v2-origid: 2269
    purely_exploratory: false
  pair_type: intent
  pairs:
    truth_lie:
      pos_desc: are truthful testimony in criminal trials
      neg_desc: are deceptive testimony in criminal trials
      pos_class:
      - truth
      neg_class:
      - lie
      example_hypotheses:
      - admits to not knowing
      - does not claim innocence
      - expresses more uncertainty
      - is more apologetic
  flip: true
tweet_gender:
- generation: whether the Twitter user is male or female
  applications:
  - target: the specific topics that each gender tends to talk about
    user: a gender studies researcher
    example_hypotheses:
    - mentions or references to outdoor activities, such as camping, hiking, and fishing
    - discusses technology and the internet
    - references literature, such as quotes from books and poems
    v2-origid: 270
    purely_exploratory: false
  - target: the speaking style of each gender
    user: a gender studies researcher
    example_hypotheses:
    - uses technical jargon
    - uses aggressive language
    - uses more slang words
    v2-origid: 271
    purely_exploratory: false
  pair_type: demographic
  pairs:
    male_female:
      pos_desc: are Tweets from male users
      neg_desc: are Tweets from female users
      pos_class:
      - male_tweets
      neg_class:
      - female_tweets
      example_hypotheses:
      - talks about sports
      - talks about politics
      - talks about celebrities
  flip: true
tweet_rumor:
- generation: how long it's been since the rumor started
  applications:
  - target: the general tone of rumors as they evolve over time
    user: a sociologist studying the patterns of rumors
    example_hypotheses:
    - uses language that is skeptical of the rumor
    - is positive in tone and offers praise
    - includes more disbelief and satire
    v2-origid: 274
    purely_exploratory: false
  pair_type: time
  pairs:
    denzel_wash_early_late:
      pos_desc: are early Twitter rumors about Denzel Washington praising Trump
      neg_desc: are later Twitter rumors about Denzel Washington praising Trump
      pos_class:
      - denzel_washington_early
      neg_class:
      - denzel_washington_late
      example_hypotheses:
      - is less skeptical
      - expresses more surprise
    redhawks_early_late:
      pos_desc: are early Twitter rumors about the Redhawks
      neg_desc: are later Twitter rumors about the Redhawks
      pos_class:
      - redhawks_early
      neg_class:
      - redhawks_late
      example_hypotheses:
      - is less skeptical
      - expresses more surprise
    veggietales_early_late:
      pos_desc: are early Twitter rumors about a Veggietales cannabis character
      neg_desc: are later Twitter rumors about a Veggietales cannabis character
      pos_class:
      - veggietales_early
      neg_class:
      - veggietales_late
      example_hypotheses:
      - is less skeptical
      - expresses more surprise
    zucc_early_late:
      pos_desc: are early Twitter rumors about Zuckerberg buying a yatch
      neg_desc: are later Twitter rumors about Zuckerberg buying a yatch
      pos_class:
      - zuckerberg_yatch_early
      neg_class:
      - zuckerberg_yatch_late
      example_hypotheses:
      - is less skeptical
      - expresses more surprise
  flip: true
twitter_bots:
- generation: the type of user posting the Tweet
  applications:
  - target: the specific topics discussed by different kinds of users, so I can detect
      bots
    user: an engineer at a social media company building a spam detector
    example_hypotheses:
    - mentions celebrities or famous people
    - contains references to sports and athletes
    - contains references to current events or news stories
    v2-origid: 277
    purely_exploratory: false
  pair_type: demographic
  pairs:
    human:
      pos_desc: were Tweeted by bots
      neg_desc: were Tweeted by humans
      pos_class:
      - trad_bot
      neg_class:
      - human
      example_hypotheses:
      - talks about sexual topics
      - discusses politics
    trad_social:
      pos_desc: were Tweeted by bots that pretend to be people
      neg_desc: were Tweeted by weaker bots
      pos_class:
      - social_bot
      neg_class:
      - trad_bot
      example_hypotheses:
      - talks about other people
  flip: true
twitter_misspellings:
- generation: whether the Tweet has a misspelling
  applications:
  - target: the social context in which people are more likely to use a misspelled
      word
    user: a linguist studying English vernacular
    example_hypotheses:
    - refers to the other person as a friend
    - is talking to a family member
    - mentions people in a casual manner, such as calling someone 'hun' or 'missy'
    v2-origid: 281
    purely_exploratory: false
  pair_type: grammar
  pairs:
    going_goin:
      pos_desc: misspell "going"
      neg_desc: don't misspell "going"
      pos_class:
      - going_misspell
      neg_class:
      - going_proper
      example_hypotheses:
      - is part of a conversation
      - is part of an argument
    that_dat:
      pos_desc: misspell "that"
      neg_desc: don't misspell "that"
      pos_class:
      - that_misspell
      neg_class:
      - that_proper
      example_hypotheses:
      - is part of a conversation
      - is more friendly
    with_wit:
      pos_desc: misspell "with"
      neg_desc: don't misspell "with"
      pos_class:
      - with_misspell
      neg_class:
      - with_proper
      example_hypotheses:
      - is part of a conversation
      - is more friendly
    your_ur:
      pos_desc: misspell "your"
      neg_desc: don't misspell "your"
      pos_class:
      - your_misspell
      neg_class:
      - your_proper
      example_hypotheses:
      - is part of a conversation
      - is more friendly
  flip: true
un_debates:
- generation: the year the speech was given
  applications:
  - target: the specific policy priorities and stances of countries over time
    user: a political scientist studying historical trends
    example_hypotheses:
    - discusses the need for a two-state solution between Israel and Palestine
    - discusses the importance of holding China accountable
    - mentions the importance of strong families and faith
    v2-origid: 287
    purely_exploratory: false
  pair_type: time
  pairs:
    russia_2k_08_12:
      pos_desc: are speeches from Russia between 2008 and 2012
      neg_desc: are speeches from Russia between 2000 and 2008
      pos_class:
      - russia_08_12
      neg_class:
      - russia_2k_08
      example_hypotheses:
      - is more friendly towards the West
      - describes economic problems
    russia_08_12_post:
      pos_desc: are speeches from Russia between 2008 and 2012
      neg_desc: are speeches from Russia after 2012
      pos_class:
      - russia_08_12
      neg_class:
      - russia_post_12
      example_hypotheses:
      - is more friendly towards the West
      - mentions military disputes
    china_2k_13_16:
      pos_desc: are speeches from China between 2000 and 2013
      neg_desc: are speeches from China between 2013 and 2016
      pos_class:
      - china_2k_13
      neg_class:
      - china_13_16
      example_hypotheses:
      - is more friendly towards the West
      - describes economic issues
    china_13_16_post:
      pos_desc: are speeches from China between 2013 and 2016
      neg_desc: are speeches from China after 2016
      pos_class:
      - china_13_16
      neg_class:
      - china_post_16
      example_hypotheses:
      - is more friendly towards the West
      - does not mention the military
    us_bush_obama:
      pos_desc: are speeches from the United States between 2001 and 2008
      neg_desc: are speeches from the United States between 2009 and 2016
      pos_class:
      - us_bush
      neg_class:
      - us_obama
      example_hypotheses:
      - talks more about Iraq
      - mentions counterterrorism
      - talks about 9/11
    us_obama_trump:
      pos_desc: are speeches from the United States between 2009 and 2016
      neg_desc: are speeches from the United States after 2017
      pos_class:
      - us_obama
      neg_class:
      - us_trump
      example_hypotheses:
      - are more friendly
      - mentions collaboration
  flip: true
- generation: the country giving the speech
  applications:
  - target: the specific policy priorities and stances of each country
    user: a political scientist studying comparative government
    example_hypotheses:
    - calls for the enforcement of compliance by states known to sponsor terrorism
    - suggests steps to enable developing countries to coordinate their development
      efforts
    - mentions measures to address climate change based on mitigation and adaptation
      strategies
    v2-origid: 298
    purely_exploratory: false
  pair_type: authorship
  pairs:
    israel_arab_80s:
      pos_desc: are speeches from Israel in the 80s
      neg_desc: are speeches from members of the Arab League in the 80s
      pos_class:
      - israel_80s
      neg_class:
      - pales_80s
      - jordan_80s
      - lebanon_80s
      - iraq_80s
      - syria_80s
      - sudan_80s
      example_hypotheses:
      - mentions Israel
      - talks about sovereignty
    israel_arab_90s:
      pos_desc: are speeches from Israel in the 90s
      neg_desc: are speeches from members of the Arab League in the 90s
      pos_class:
      - israel_90s
      neg_class:
      - pales_90s
      - jordan_90s
      - lebanon_90s
      - iraq_90s
      - syria_90s
      - sudan_90s
      example_hypotheses:
      - supportive of Israel
      - talks about sovereignty
      - mentions anti-semitism
    israel_arab_2ks:
      pos_desc: are speeches from Israel from 2000-2009
      neg_desc: are speeches from members of the Arab League from 2000-2009
      pos_class:
      - israel_2ks
      neg_class:
      - pales_2ks
      - jordan_2ks
      - lebanon_2ks
      - iraq_2ks
      - syria_2ks
      - sudan_2ks
      example_hypotheses:
      - does not mention nuclear weapons
      - mentions anti-semitism
    israel_arab_10s:
      pos_desc: are speeches from Israel in the 2010s
      neg_desc: are speeches from members of the Arab League in the 2010s
      pos_class:
      - israel_10s
      neg_class:
      - pales_10s
      - jordan_10s
      - lebanon_10s
      - iraq_10s
      - syria_10s
      - sudan_10s
      example_hypotheses:
      - supportive of Israel
      - mentions anti-semitism
    israel_gulf_80s:
      pos_desc: are speeches from Israel in the 80s
      neg_desc: are speeches from members of the Gulf Cooperation Council in the 80s
      pos_class:
      - israel_80s
      neg_class:
      - bahrain_80s
      - kuwait_80s
      - oman_80s
      - qatar_80s
      - saudi_80s
      - uae_80s
      example_hypotheses:
      - mentions Israel
      - mentions anti-semitism
    israel_gulf_90s:
      pos_desc: are speeches from Israel in the 90s
      neg_desc: are speeches from members of the Gulf Cooperation Council in the 90s
      pos_class:
      - israel_90s
      neg_class:
      - bahrain_90s
      - kuwait_90s
      - oman_90s
      - qatar_90s
      - saudi_90s
      - uae_90s
      example_hypotheses:
      - supportive of Israel
      - mentions anti-semitism
    israel_gulf_2ks:
      pos_desc: are speeches from Israel from 2000-2009
      neg_desc: are speeches from members of the Gulf Cooperation Council from 2000-2009
      pos_class:
      - israel_2ks
      neg_class:
      - bahrain_2ks
      - kuwait_2ks
      - oman_2ks
      - qatar_2ks
      - saudi_2ks
      - uae_2ks
      example_hypotheses:
      - does not mention nuclear weapons
      - mentions anti-semitism
    israel_gulf_10s:
      pos_desc: are speeches from Israel in the 2010s
      neg_desc: are speeches from members of the Gulf Cooperation Council in the 2010s
      pos_class:
      - israel_10s
      neg_class:
      - bahrain_10s
      - kuwait_10s
      - oman_10s
      - qatar_10s
      - saudi_10s
      - uae_10s
      example_hypotheses:
      - talks about Palestine
      - talks about attacks
    indopak_80s:
      pos_desc: are speeches from India in the 80s
      neg_desc: are speeches from Pakistan in the 80s
      pos_class:
      - india_80s
      neg_class:
      - pak_80s
      example_hypotheses:
      - is hostile towards Pakistan
      - mentions military disputes
    indopak_90s:
      pos_desc: are speeches from India in the 90s
      neg_desc: are speeches from Pakistan in the 90s
      pos_class:
      - india_90s
      neg_class:
      - pak_90s
      example_hypotheses:
      - talks about border conflicts
      - talks about Muslims
    indopak_2ks:
      pos_desc: are speeches from India from 2000-2009
      neg_desc: are speeches from Pakistan from 2000-2009
      pos_class:
      - india_2ks
      neg_class:
      - pak_2ks
      example_hypotheses:
      - mentions economic issues
      - talks about China
    indopak_10s:
      pos_desc: are speeches from India in the 2010s
      neg_desc: are speeches from Pakistan in the 2010s
      pos_class:
      - india_10s
      neg_class:
      - pak_10s
      example_hypotheses:
      - is friendly towards the West
      - mentions democracy
    korea_japan_90s:
      pos_desc: are speeches from Korea in the 90s
      neg_desc: are speeches from Japan in the 90s
      pos_class:
      - korea_90s
      neg_class:
      - japan_90s
      example_hypotheses:
      - is hostile towards Japan
      - mentions economic issues
    korea_japan_2ks:
      pos_desc: are speeches from Korea from 2000-2009
      neg_desc: are speeches from Japan from 2000-2009
      pos_class:
      - korea_2ks
      neg_class:
      - japan_2ks
      example_hypotheses:
      - mentions the technology sector
      - mentions North Korea
    korea_japan_10s:
      pos_desc: are speeches from Korea in the 2010s
      neg_desc: are speeches from Japan in the 2010s
      pos_class:
      - korea_10s
      neg_class:
      - japan_10s
      example_hypotheses:
      - mentions China
      - mentions North Korea
  flip: true
unhealthy_conversations:
- generation: the type of unhealthy tone as annotated by experts
  applications:
  - target: what specific topics and actions define each tone
    user: a couples therapist learning about different unhealthy conversation styles
    example_hypotheses:
    - refers to other people as 'twerps' or 'idiots'
    - refers to other people as 'stupid' or 'horrible'
    - mentions intervening in a situation
    v2-origid: 305
    purely_exploratory: false
  pair_type: judgement
  pairs:
    dismissive_condescending:
      pos_desc: are online messages that are dismissive
      neg_desc: are online messages that are condescending
      pos_class:
      - dismissive
      neg_class:
      - condescending
      example_hypotheses:
      - is not insulting
      - does not elaborate
    generalisation_unfair:
      pos_desc: are online messages that generalize
      neg_desc: are online messages that generalize unfairly
      pos_class:
      - generalisation
      neg_class:
      - generalisation_unfair
      example_hypotheses:
      - is more objective
      - provides more reasoning
    hostile_antagonize:
      pos_desc: are online messages that are hostile
      neg_desc: are online messages that are antagonizing
      pos_class:
      - hostile
      neg_class:
      - antagonize
      example_hypotheses:
      - is less insulting
      - uses more curse words
  flip: true
urban_dictionary:
- generation: how many upvotes or downvotes the definition received
  applications:
  - target: what types of proposed definitions people like
    user: a user of the site hoping to write popular definitions
    example_hypotheses:
    - uses slang terms
    - mentions celebrities or popular figures
    - presents an aspirational or upbeat message
    v2-origid: 308
    purely_exploratory: false
  pair_type: judgement
  pairs:
    top1_down:
      pos_desc: are Urban Dictionary definitions in the top 1% of downvotes
      neg_desc: are average Urban Dictionary definitions
      pos_class:
      - top_1_down
      neg_class:
      - control
      example_hypotheses:
      - is racist
      - is sexist
      - is incoherent
      - is not funny
    top1_top5_down:
      pos_desc: are Urban Dictionary definitions in the top 5% of upvotes
      neg_desc: are Urban Dictionary definitions in the top 10% of upvotes
      pos_class:
      - top_1_down
      - top_5_down
      neg_class:
      - top_10_down
      example_hypotheses:
      - is more controversial
    top1_top5_up:
      pos_desc: are Urban Dictionary definitions in the top 1% of upvotes
      neg_desc: are Urban Dictionary definitions in the top 5% of upvotes
      pos_class:
      - top_1_up
      neg_class:
      - top_5_up
      example_hypotheses:
      - is about a very common word
    top1_up:
      pos_desc: are Urban Dictionary definitions in the top 1% of upvotes
      neg_desc: are average Urban Dictionary definitions
      pos_class:
      - top_1_up
      neg_class:
      - control
      example_hypotheses:
      - includes a joke
      - describes a common word
    top5_top10_up:
      pos_desc: are Urban Dictionary definitions in the top 5% of upvotes
      neg_desc: are Urban Dictionary definitions in the top 10% of upvotes
      pos_class:
      - top_1_up
      - top_5_up
      neg_class:
      - top_10_up
      example_hypotheses:
      - is about a very common word
      - provides a short definition
  flip: true
yc_startups:
- generation: where the startup was founded
  applications:
  - target: how location influences the service or product offered
    user: an aspiring entreprenuer deciding whether to move to the bay
    example_hypotheses:
    - references the use of software or applications
    - aims to reduce time and/or cost of a process
    - uses data analytics to inform decisions
    v2-origid: 311
    purely_exploratory: false
  pair_type: location
  pairs:
    bay_area:
      pos_desc: are Y Combinator startup descriptions from the United States
      neg_desc: are Y Combinator startup descriptions outside the United States
      pos_class:
      - location_usa
      neg_class:
      - location_not_usa
      example_hypotheses:
      - describes a business-to-business company
      - talks about machine learning
  flip: true
- generation: the operation status of the startup
  applications:
  - target: what services or products are more likely to succeed
    user: a venture capital firm deciding which startups to fund
    example_hypotheses:
    - offers a cloud-based service
    - enables users to manage or track tasks online
    - enables users to access cutting edge DNA sequencing technology
    v2-origid: 313
    purely_exploratory: false
  pair_type: subject
  pairs:
    dead_operating:
      pos_desc: are Y Combinator startup descriptions that are inactive
      neg_desc: are Y Combinator startup descriptions that are still active
      pos_class:
      - status_inactive
      neg_class:
      - status_active
      example_hypotheses:
      - is part of an existing product
      - mentions consumer goods
    exited_operating:
      pos_desc: are Y Combinator startup descriptions that have exited through acquisition or IPO
      neg_desc: are Y Combinator startup descriptions that are still operating
      pos_class:
      - status_acquired
      - status_public
      neg_class:
      - status_active
      example_hypotheses:
      - is a piece of technology
      - mentions machine learning
      - is about databases
  flip: true
- generation: when the startup was founded
  applications:
  - target: which products and services are most exciting today
    user: a venture capital firm studying trends in the industry
    example_hypotheses:
    - involves gesture control technology
    - mentions synthetic biology
    - mention artificial intelligence or machine learning
    v2-origid: 314
    purely_exploratory: false
  pair_type: time
  flip: true
  pairs:
    years_0609_1013:
      pos_desc: are Y Combinator startup descriptions between 2006 and 2009
      neg_desc: are Y Combinator startup descriptions between 2010 and 2013
      pos_class:
      - year_06
      - year_07
      - year_08
      - year_09
      neg_class:
      - year_10
      - year_11
      - year_12
      - year_13
      example_hypotheses:
      - talks about the internet
      - does not mention machine learning
    years_1013_1417:
      pos_desc: are Y Combinator startup descriptions between 2010 and 2013
      neg_desc: are Y Combinator startup descriptions between 2014 and 2017
      pos_class:
      - year_10
      - year_11
      - year_12
      - year_13
      neg_class:
      - year_14
      - year_15
      - year_16
      - year_17
      example_hypotheses:
      - talks about the internet
      - does not mention machine learning
    years_1417_1822:
      pos_desc: are Y Combinator startup descriptions between 2014 and 2017
      neg_desc: are Y Combinator startup descriptions between 2018 and 2022
      pos_class:
      - year_14
      - year_15
      - year_16
      - year_17
      neg_class:
      - year_18
      - year_19
      - year_20
      - year_21
      - year_22
      example_hypotheses:
      - talks about the internet
      - does not mention machine learning
  flip: true
- generation: the identity of the founder
  applications:
  - target: how founder diversity affects the product or service offered
    user: a venture capital firm studying trends in the industry
    example_hypotheses:
    - provides services for underprivleged groups
    - incorporates a community element in the product
    - is not driven by profit motive 
    v2-origid: 316
    purely_exploratory: false
  pair_type: demographic
  flip: true
  pairs:
    highlight_black:
      pos_desc: are Y Combinator startups with black founders
      neg_desc: are Y Combinator startups with nonblack founders
      pos_class:
      - highlight_black_yes
      neg_class:
      - highlight_black_no
    highlight_latinx:
      pos_desc: are Y Combinator startups with Latinx founders
      neg_desc: are Y Combinator startups with non-Latinx founders
      pos_class:
      - highlight_latinx_yes
      neg_class:
      - highlight_latinx_no
    highlight_women:
      pos_desc: are Y Combinator startups with female founders
      neg_desc: are Y Combinator startups with male founders
      pos_class:
      - highlight_women_yes
      neg_class:
      - highlight_women_no