Wikidata talk:WikiProject European Union

From Wikidata
Jump to navigation Jump to search
 Main Page Project talk Data models Properties 


[edit]

I have been working on building a data model for Wikidata items about EU legal acts (regulations, directives etc.) and writing SPARQL queries for pulling data out of Cellar SPARQL interface (CC0 licenced). I did the first test batch for existing Wikidata items about EU legal acts that have been passed between 2020 and today. A few issues have risen, and they need to be addressed before doing any larger batches or creating new items:

  1. Labels. If an act doesn't already have a label, what should it be? "Regulation (EU) XXXX/XXX of the European Parliament and of the Council", "Regulation (EU) XXXX/XXX", "Regulation XXXX/XXX"? What about in other languages?
  2. There's around 6000 EU legal acts where official name (P1448) is used. I think title (P1476) should be used instead.
  3. Currently legal citation of this text (P1031) has mostly falsely formatted values. Another problem is that the correct format differs language to language, and it is not easily available through the Cellar. Easiest solution would be to skip the property, and delete the existing value. ✓ Done
  4. Should law identifier (P8550) be used? If yes, how? Maybe just the numbers, e.g. "2016/679" for GDPR.
  5. Should we use country (P17) = European Union (Q458) on legal acts or only the property applies to jurisdiction (P1001)?
  6. Should we use published in (P1433) = Official Journal of the European Union, L series (Q26284397) or Official Journal of the European Union (Q480448)?
  7. Some of the legal act types have little used subclasses that might be good to merge into the parent class, e.g. Commission Regulation (Q38126860), regulation of the European Parliament and of the Council (Q38126804). See all here.

If you have any thoughts, feel free to join the discussion. Samoasambia ✎ 21:32, 27 September 2024 (UTC)[reply]

I just removed the incorrect legal citations with this QS batch. Samoasambia ✎ 15:28, 28 September 2024 (UTC)[reply]
The second issue was solved with these batches: [1], [2] + a few small ones. Around 200 titles had double whitespaces in them, so fixed them as well. Samoasambia ✎ 16:59, 4 October 2024 (UTC)[reply]

Adding all titles of CELEX documents

[edit]

I created Wikidata:Requests for permissions/Bot/So9qBot 8 to import names by scraping them. CELLAR did not have the titles according to my research back when I wrote the code. So9q (talk) 09:38, 7 October 2024 (UTC)[reply]

The documentation page of Cellar is a monstrosity but I managed to pull out titles from it. The SPARQL code and the link to the endpoint is available on the Data models subpage inside one of the collapsable boxes. However, scraping titles is completely fine for me, it makes the job slightly easier. Samoasambia ✎ 22:22, 8 October 2024 (UTC)[reply]

Missing property for end of force dates

[edit]

See https://www.wikidata.org/wiki/Wikidata_talk:WikiProject_Law#Which_date_property_should_I_use_for_EU_law_no_longer_in_force%3F I just checked and it seems it is still missing. So9q (talk) 09:39, 7 October 2024 (UTC)[reply]

Creating an opposite property for effective date (P7588) is a good idea. Currently repealed by (P2568) has a required qualifier point in time (P585) which tells the end of force date but some acts may have sunset provision (Q779351) which means there is no repealing act that causes the ending. On EUR-Lex acts seem usually to come into force at midnight, which means that the "date of end of validity" of the repealed act is always the previous day (e.g. 31 December 2023) while the "date of effect" of the new act is the next day (e.g. 1 January 2024). In that case the value of the property should probably be effective date (P7588) minus 1 day? Samoasambia ✎ 22:02, 7 October 2024 (UTC)[reply]

Corporate body classification

[edit]
Samoasambia Julius Schwarz User:So9q

Notified participants of WikiProject European Union: Hi friends and happy December! :)

While we are waiting for the EU Corporate body code property to be created, I went to look how we should classify all the EU institutions, bodies, agencies, offices etc. Currently on Wikidata the top class is body of the European Union (Q4936585) and it has several subclasses. But actually the EU uses the word "body" to refer to only 8 bodies listed here (i.e. EEAS, EESC, COR, EIB, Ombudsman, EDPS, EDPB and EPPO) and the other EU institutions, agencies, offices etc. are not subclasses of it.

I found that the Publications Office has published a Corporate body classification that is meant to accompany the Corporate body list. Click the "Tree view" tab and you will see the class structure; I also put it onto this subpage together with descriptions and matching Wikidata items. The classification has "European Union corporate body" as its top class, and I think the corresponding Wikidata item for it would be corporate body of European Union (Q112861866). Based on this official classification I have made a proposal for a new class structure to be implemented here on Wikidata. I put it in the box below to save some space:

Institutional ontology proposal

Sources: [3][4][5]

Feel free to comment! Regards Samoasambia ✎ 15:53, 2 December 2024 (UTC)[reply]

LGTM! Thanks for taking the time to work on this :) So9q (talk) 17:24, 2 December 2024 (UTC)[reply]
Retrieved from "https://www.wikidata.org/w/index.php?title=Wikidata_talk:WikiProject_European_Union&oldid=2282770861"