User:Hall1467/Determining Important Properties for Recommendation

From Meta, a Wikimedia project coordination wiki
Created
June 5, 2017
Contact
Andrew Hall
Collaborators
Aaron Halfaker
Duration:  2017-07 – 2017-09

This page documents a proposed research project.
Information may be incomplete and may change before the project starts.


Starting in July, we would like funding to continue our current project which is seeking to explore the usage and value of Wikidata properties, statements, and entities. This work has important implications for Wikidata and its contributors. For example, our findings can help improve Wikidata's property suggester by recommending valuable properties/statements that do not exist. In fact, this in particular is one primary goals of our project. With this project, we could additionally also provide other benefits to Wikidata contributors such as entity recommendations; valuable client pages that don't have an item could be recommended.

In our work, we'd like to deeply explore how Wikidata property/statement/entity contributions align with valuable content in clients. For example, we may find that certain types of entities or statements exist in Wikidata, but are not valuable to clients. Alternatively, we may find that there are certain types of entities or statements that do not exist, but that could be very valuable. We'd like to explore the various dimensions in which misalignments may occur and consider potential solutions to them. We would also like to take a deeper look into the characteristics of valuable properties, statements, and entities and really understand what defines them. All of this information can help create better editor tools.

While our primary focus has (so far) been on looking into Wikidata usage on client wikis, we'd also like to work towards tracking/analyzing third-party usage of Wikidata (this could possibly be done through a log point, for example).

We feel the outcome of this work will have clear value for WMF, WMDE, and Wikidata. For some more details, please see our original WMF proposal: Research:Understanding Wikidata's Value.