Search improvements

[ntarocco]

We need to improve the search results in InvenioRDM. Currently, the search mappings are very basic and tuned to take into account different search queries that users might input.

See also inveniosoftware/invenio-vocabularies#335
See CERN Search mappings tuning: https://gitlab.cern.ch/webservices/cern-search/cern-search-rest-api

A related issue already exists, with a PR partially addressing this problem.

Scope

Improve searches of:

• records
• communities
• users
• groups
• vocabularies (affiliations, grants, subjects, names, licenses....)

Typical search queries

• input of accents, apostrophes, other unicode characters
• partial input: searching for Mar, should already propose results for Mario without typing the entire name

Known issues

• Users are findable only when the exact name or e-mail is inserted
• Searching for Lopez Group does not show the exact record.

Open questions

• What do we do with titles with LaTeX inside? Do we completely remove the LaTeX formula and it is not searchable?

Tasks

Preview Give feedback

1. Improve diacritic search #2746
   +0
2. Improve exact keyword search #2747
   +0
3. Some users cannot be found when inviting or setting restrictions #2752
   +0
4. Searching for names when adding creators return inconsistent results #2756
   +0
5. Searching for prefixes of words fails #2759
   +0
6. Searching with special characters result to 500 in affiliations #2761
   +0
7. creators search: "Something went wrong" error when entering a few letters in search #1849
   bug +1
8. Search fails when forward slashes are used #2766
   +0

Search issues not related to search results

Tasks

Preview Give feedback

1. landing page: search link for contributors is wrong #2247
   bug +1
2. search-ui: consistent comma separators for large numbers #2051
   stale ux +2
3. https://github.com/zenodo/rdm-project/issues/618

[github-actions]

This issue was automatically marked as stale.