Page MenuHomePhabricator

Integrate Wikidata statements to our MR API
Closed, ResolvedPublic

Description

As a customer, when I access the MR API endpoint, I want the JSON to include the WIkidata statements for the related page entity

Subtask As a first step to speed up Wikidata API, I'll investigate caching of P___ property labels and descriptions in EN (maybe more languages) so we can avoid multiple trips to WIkidata APIs for labels.

I'll parser this HTML page to collate the PID and labels in a Golang map (or Redis cache): https://www.wikidata.org/wiki/Wikidata:Database_reports/List_of_properties/all

  • Add the Wikidata entity statements to the MR API
  • Add filtering so the statements can be removed from the API
  • Run a performance test to see how long it takes for the new JSON to be sent for one REST request

Event Timeline

After the investigation of Wikidata statements, the conclusion is that a more domain-specific approach is needed. Plain copying of the statements is not what our clients are looking for. They want prebuilt knowledge graphs for clusters of wikipedia/wikidata articles: Music, TV, Movies, Sports, and Celebrity searches, where we can provide sub-trees of knowledge. To do this we need a list of sub-domains and search examples to scope out the requirements for a simplified knowledge graph.

JArguello-WMF claimed this task.