Global search shows incorrect results when a uri is defined in many vocabularies

[henriyli]

At which URL did you encounter the problem?

http://finto.fi/en/search?anylang=on&q=vesi-+ja+ymp%C3%A4rist%C3%B6piirit*

What steps will reproduce the problem?

1. Open the provided link
2. Look at the broader concepts of the first result ( AFO: vesi- ja ympäristöpiirit (hallintoalueet) (fi) )
3. Look at the broader concept on the actual concept page:
   http://finto.fi/afo/en/page/?clang=fi&uri=http%3A%2F%2Fwww.yso.fi%2Fonto%2Fyso%2Fp21720

What is the expected output? What do you see instead?

Expected to only see the broader concept that is defined in AFO ("hallinnollinen piiri (hallintoalueet)"). Instead I see two concepts "hallinnolliset piirit" and "hallintoalueet", first of which is the correct concept but is presented with an incorrect prefLabel from some other vocabulary. This happens because different values are provided for the same URI in different ontologies.

There are two related issues here:

1. The values for skos:broader are incorrect.
2. The skos:prefLabels of those concepts are incorrect.

This leads to a nasty situation, where the search results are inaccurate for both the current YSO and the somewhat outdated domain ontologies.

[tfrancart]

I need to understand this because I may have a similar situation : is it because the same concept is in 2 different vocabularies in Skosmos (2 different vocabularies defined in vocabularies.ttl, pointing to the same triplestore but 2 different named graphs) ? Thanks for the clarification

[osma]

@tfrancart Exactly. This happens because the AFO data (in one named graph) includes an old version of YSO, so it has the old labels for that concept URI. Another graph on the same triple store contains the current YSO.

The problem is that the search result page performs a CONSTRUCT query that takes information about all the concepts on that page, from all named graphs on the same endpoint, and the triples get mixed up in the constructed graph.

Fixing this would require another approach to constructing search result pages. But it's quite difficult to do this in a way that has reasonable performance.

[tfrancart]

Many thanks. Just to let you know my use-case : I want to publish all the historical versions of one or more vocabularies in Skosmos, to allow users to choose the version they want to see, and eventually jump from one version to another ("See this concept in V1, V2, V3"). So I will definitely run in the same issue. I will provide more details on my use-case in a separate issue later. 2017-09-29 9:55 GMT+02:00 Osma Suominen <[email protected]>:

…

@tfrancart <https://github.com/tfrancart> Exactly. This happens because the AFO data (in one named graph) includes an old version of YSO, so it has the old labels for that concept URI. Another graph on the same triple store contains the current YSO. The problem is that the search result page performs a CONSTRUCT query that takes information about all the concepts on that page, from all named graphs on the same endpoint, and the triples get mixed up in the constructed graph. Fixing this would require another approach to constructing search result pages. But it's quite difficult to do this in a way that has reasonable performance. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#611 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ACmj8es04iQ33b2V3kWK6dH5Em6JooQiks5snKJVgaJpZM4NKT8Y> .

-- *Thomas Francart* -* SPARNA* Web de *données* | Architecture de l'*information* | Accès aux *connaissances* blog : blog.sparna.fr, site : sparna.fr, linkedin : fr.linkedin.com/in/thomasfrancart tel : +33 (0)6.71.11.25.97, skype : francartthomas

[osma]

@tfrancart Are you using skos-history by any chance? Since it stores the different versions of a vocabulary in different named graphs, you could probably use it together with Skosmos the way you described. But of course it is mainly useful if you want to use SPARQL to compare different versions, for example to find new concepts.

[tfrancart]

No, I don't use skos-history. We are not going as far as documenting semi-automatically the changes between versions. Just publishing properly each versions will be enough work as a first step :-) 2017-09-29 12:18 GMT+02:00 Osma Suominen <[email protected]>:

…

@tfrancart <https://github.com/tfrancart> Are you using skos-history by any chance? Since it stores the different versions of a vocabulary in different named graphs, you could probably use it together with Skosmos the way you described. But of course it is mainly useful if you want to use SPARQL to compare different versions, for example to find new concepts. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#611 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ACmj8crxYM523b4R7E1Kq79O4AQDfCS3ks5snMQJgaJpZM4NKT8Y> .

-- *Thomas Francart* -* SPARNA* Web de *données* | Architecture de l'*information* | Accès aux *connaissances* blog : blog.sparna.fr, site : sparna.fr, linkedin : fr.linkedin.com/in/thomasfrancart tel : +33 (0)6.71.11.25.97, skype : francartthomas

[osma]

Closing and archiving issue due to lack of activity. Please reopen if there are concrete reasons for reconsidering.