Page MenuHomePhabricator

Improve the inter-lingual section recommender system
Closed, ResolvedPublic

Description

In the context of the Knowledge Gaps project, we have developed a Section Recommender System to detect missing sections on articles, based on inter-lingual information.

Improvements:

  • Add verbose mode to explain where recommendations come from.
  • Improve the recommendation on stubs
  • Fixed APIs Json format issues.

Documentation about the new features can be found here: https://meta.wikimedia.org/wiki/Research:Expanding_Wikipedia_articles_across_languages/Inter_language_approach#Section_Recommendation

Event Timeline

@diego can you make the parent of this task explicit using one of the outputs 1-5 listed at the address-knowledge-gaps board?

The documentation for the updated version of the section recommender system can be found here:
https://meta.wikimedia.org/wiki/Research:Expanding_Wikipedia_articles_across_languages/Inter_language_approach#Section_Recommendation

Features added:

  • Verbose mode with contextual information (parameter verbose=True)
  • useMoreLike: implement the search on similar Wikidata items when there no sections to recommend. For example, if is not possible to find recommendations for en:Pythonides (Q1760915) article, this feature will allow to search for similar Wikidata items using the cirrus morelike api, and using them as seed for providing recommendations. Note that seed articles can come for any of the supported languages. The seed language is defined by the auxLang parameter. (parameter verbose=True)

Fixed JSON format issues for the APIs. Now they are working correctly.

https://secrec.wmflabs.org/API/recommendation/en/Quilombo