Research talk:Understanding Wikidata's Value

Add topic
From Meta, a Wikimedia project coordination wiki
Latest comment: 6 years ago by Hall1467 in topic WDQS
See the /Archive for old threads

Google versus external identifiers[edit]

Your first mention of researching value to others is Google. More obvious, from a Wikidata point of view, is the relation with the organisations we obviously collaborate with. Regularly identifiers to all kinds of organisations are announced and you find as a result identifiers to a rich set of external "Sources". Already identifiers to one "Source" like the Getty are used to find identifiers to VIAF. People at the OCLC acknowledge that having VIAF and WorldCat included in the {{authority control}} drives people to go to WorldCat and is getting people to libraries. OCLC moved from linking to English Wikipedia to Wikidata because it links to all Wikipedias. We do not know what effect it had. We do not know how to increase the number of people who do not speak English to their libraries.

Data from the live sciences are finding their way to Wikidata. Data from SwissProt among others. We do not know what its effect is and how this can be increased. We know that a lot of GLAM data is in Commons. We have told them to keep their meta data for the day when Commons gets wikidatified. We have created "Creator" templates and they now get their data from Wikidata.

There are a few identifiers to Google. I do not know their value to us or to Google. What I do know is that Google is like any other external source we connect to. It is relevant to understand what links work and what they do. Not only for them but also for us. We do not document the stories behind these collaborations. Yes research is welcome. Thanks, GerardM (talk) 05:44, 29 April 2017 (UTC)Reply

@GerardM: Thank you very much for the comment. It was both interesting and informative. Hall1467 (talk) 18:13, 31 May 2017 (UTC)Reply


We are a few that are interested in the Wikidata Query Service (WDQS) and its ability to assemble data across items. I am using it for Scholia ( WDQS allows you to determine how third-parties or second-and-a-half-parties are using Wikidata data. So you might want to get the Blazegraph/WDQS log if it exists (see also grafana [1] [2].

Note that some information on Wikipedias can be constructed from Wikidata without involving direct update. Magnus Manske's Listeria is an obvious candidate. My Scholia can also create a biography based on Wikidata, see w:Robert Kelly (political analyst)#Select bibliography.

Finn Årup Nielsen (fnielsen) (talk) 12:51, 1 May 2017 (UTC)Reply

@Fnielsen: Thanks very much for the comment, and my apologies for the slow response. Related to Scholia, you said that Wikidata use is possible "without involving direct update." Could you point me to some specifics about how this works? Is Wikidata accessed through a Lua module? Again, thanks for the comment and nice meeting you :) Hall1467 (talk) 20:50, 1 June 2017 (UTC)Reply

Goal and approach[edit]

Hello Andrew, thanks for the interesting proposal. I am looking forward to see the results. My two cents: It would be helpful, I guess, if you explain in the introduction a little more about what you want to find out, and what is the general way how to want to find it out (the general approach, e.g. technical vs. social research). For example, if I'd conduct this research I'd look specifically at the cultural and social context of the use, and ask involved volunteers of the Wikimedia (or other) websites in question. I understand that your approach is more coming from the computer sciences. Maybe you could be more explicit. Kind regards Ziko (talk) 21:31, 1 May 2017 (UTC)Reply

@Ziko: Thanks very much for the comment and my apologies for the late reply. To hopefully clear things up some, our planned approach is technical, quantitative data analysis. However, its possible that our results from this analysis will open the door to new questions. For example, why did we observe a particular result? To answer why, it would be valuable to interview and/or survey contributors who would have insight into our quantitative findings. A mixed methods approach like this could lead to deeper insights. Thanks again for the comment. Hall1467 (talk) 18:44, 31 May 2017 (UTC)Reply