The Wikidata scholarly profile page

Posted on Updated on

my_coauthors

Recently Lambert Heller wrote an overview piece on websites for scholarly profile pages: “What will the scholarly profile page of the future look like? Provision of metadata is enabling experimentation“. There he tabularized the features of the various online sites having scholarly profile pages. These sites include (with links to my entries): ORCID, ResearchGate, Mendeley, Pure and VIVO (don’t know these two), Google Scholar and Impactstory. One site missing from the equation is Wikidata. It can produce scholarly profile pages too. The default Wikidata editing interface may not present the data in a nice way – Magnus Manske’s Reasonator – better, but very much of the functionality is there to make a scholarly profile page.

In terms of the features listed by Heller, I will here list the possible utilization of Wikidata:

  1. Portrait picture: The P18 property can record Wikimedia Commons image related to a researcher. For instance, you can see a nice photo of neuroimaging professor Russ Poldrack.
  2. Researchers alternative names: This is possible with the alias functionality in Wikidata. Poldrack is presently recorded with the canonical label “Russell A. Poldrack” and the alternative names “Russell A Poldrack”, “R. A. Poldrack”, “Russ Poldrack” and “R A Poldrack”. It is straightforward to add more variations
  3. IDs/profiles in other systems: There are absolutely loads of these links in Wikidata. To name a few deep linking posibilities: Twitter, Google Scholar, VIAF, ISNI, ORCID, ResearchGate, GitHub and Scopus. Wikidata is very strong in interlinking databases.
  4. Papers and similar: Papers are presented as items in Wikidata and these items can link to the author via P50. The reverse link is possible with a SPARQL query. Futhermore, on the researcher’s items it is possible to list main works with the appropriate property. Full texts can be linked with the P953 property. PDF of papers with an appropriate compatible license can be uploaded to Wikimedia Commons and/or included in Wikisource.
  5. Uncommon research product: I am not sure what this is, but the developer of software services is recorded in Wikidata. For instance, for the neuroinformatics database OpenfMRI it is specified that Poldrack is the creator. Backlinks are possible with SPARQL queries.
  6. Grants, third party funding. Well there is a sponsor property but how it should be utilized for researchers is not clear. With the property, you can specify that paper or research project were funded by an entity. For the paper The Center for Integrated Molecular Brain Imaging (Cimbi) database you can see that it is funded by the Lundbeck Foundation and Rigshospitalet.
  7. Current institution: Yes. Employer and affiliation property is there for you. You can see an example of an incomplete list of people affiliated with research sections at my department, DTU Compute, here, – automagically generated by the Magnus Manske’s Listeria tool.
  8. Former employers, education etc.: Yes. There is a property for employer and for affiliation and for education. With qualifiers you can specify the dates of employment.
  9. Self assigned keywords: Well, as a Wikidata contributor you can create new items and you can use these items for specifying field of work of to label you paper with main theme.
  10. Concept from controlled vocabulary: Whether Wikidata is a controlled vocabulary is up for discussion. Wikidata items can be linked to controlled vocabularies, e.g., Dewey’s, so there you can get some controlness. For instance, the concept “engineer” in Wikidata is linked the BNCF, NDL, GND, ROME, LCNAF, BNF and FAST.
  11. Social graph of followers/friends: No, that is really not possible on Wikidata.
  12. Social graph of coauthors: Yes, that is possible. With Jonas Kress’ work on D3 enabling graph rendering you got on-the-fly graph rendering in the Wikidata Query Service. You can see my coauthor graph here (it is wobbly at the moment, there is some D3 parameter that need a tweak).
  13. Citation/attention metadata from platform itself: No, I don’t think so. You can get page view data from somewhere on the Wikimedia sites. You can also count the number of citations on-the-fly, – to an author, to a paper, etc.
  14. Citation/attention metadata from other sources: No, not really.
  15. Comprehensive search to match/include own papers: Well, perhaps not. Or perhaps. Magnus Manske’s sourcemd and quickstatement tools allow you to copy-paste a PMID or DOI in a form field press two buttons to grap bibliographic information from PubMed and a DOI source. One-click full paper upload is not well-supported, – to my knowledge. Perhaps Daniel Mietchen knows something about this.
  16. Forums, Q&A, etc.: Well, yes and no. You can use the discussion pages on Wikidata, but these pages are perhaps mostly for discussion of editing, rather than the content of the described item. Perhaps Wikiversity could be used.
  17. Deposit own papers: You can upload appropriately licensed papers to Wikimedia Commons or perhaps Wikisource. Then you can link them from Wikidata.
  18. Research administration tools: No.
  19. Reuse of data from outside the service: You better believe! Although Wikidata is there to be used, a mass download from the Wikidata Query Service can run into timeout problems. To navigate the structure of individual Wikidata item, you need programming skills, – at least for the moment. If you are really desperate you can download the Wikidata dump and Blazegraph and try to setup your own SPARQL server.

 

13 thoughts on “The Wikidata scholarly profile page

    E.L. Willighagen said:
    October 1, 2016 at 7:42 am

    About step 4, if I use the tool by Magnus Manske (http://tools.wmflabs.org/sourcemd/) to add papers to Wikidata, it does not use P50 for authors, but P2093, e.g. as in https://www.wikidata.org/wiki/Q26899110 … So, it doesn’t link the paper to my Wikidata page. Do you know how this should be resolved? Are there multiple ways articles can be represented? Is Wikicite compatible with Magnus’ tool? Basically, do you have suggestions?

      E.L. Willighagen said:
      October 1, 2016 at 8:21 am

      I just noticed you actually P2093 with a P50 statement on one of my articles two days ago! So, that’s the way forward? That doesn’t break any existing uses?

        Finn Årup Nielsen responded:
        October 1, 2016 at 3:16 pm

        As I understand the P2093 should be regarded as a temporary property until someone move the information to P50. P2093 was suggested by Magnus Manske in connection with his work on sourcemd, see the property proposal. Any bibliographic tool using an item with P2093 and P50 needs to carefully handle both and the associated P1545 qualifier.

      Finn Årup Nielsen responded:
      October 1, 2016 at 3:39 pm

      My approach is to use sourcemd and then manually resolve to P50 for the authors that are items and erase the corresponding P2093. Sometimes I create authors and then P50 them. For conference article with no DOI and PMID one currently need to create everything manually.

      “Is Wikicite compatible with Magnus’ tool?” I do not know what you mean by that. Wikicite is “just” an idea/vision/project. I would say Magnus’ sourcemd is a Wikicite tool. It uses the properties you would expect, but let humans or other tools to do the main theme, P50-author and the P2860-citation.

      Finn Årup Nielsen responded:
      October 1, 2016 at 3:55 pm

      It may also worth to mention Magnus Manske’s gamification https://tools.wmflabs.org/wikidata-game/distributed/#game=9. I find it too difficult to use: You are mostly presented with an author you do not know and a paper outside your field, so determining whether “H. Yang” in “Search for long-lived gravitational-wave transients coincident with long gamma-ray bursts” is “Hexion Yang” is really difficult.

    Andy Mabbett said:
    October 3, 2016 at 3:43 pm

    The problem with P2093 (“author name string”) is that it is not guaranteed to be unique – the very problem that ORCID is designed to solve.

      Finn Årup Nielsen responded:
      October 3, 2016 at 4:05 pm

      Yes, a lot of “F Nielsen”, “J Larsen” and similar exist. I suppose Wikidata with P50 is also designed to solved that problem.

      ChristianKl said:
      October 3, 2016 at 4:47 pm

      If you already have the OrcID numbers you can use P50 and there’s no need for P2093. P2093 exist because sometimes we don’t know more about the author than is written in the paper.

        Finn Årup Nielsen responded:
        October 3, 2016 at 5:01 pm

        Does Magnus’ tool look for ORCID?

    Finn Årup Nielsen responded:
    October 3, 2016 at 4:10 pm
    Finn Årup Nielsen responded:
    October 21, 2016 at 2:55 pm

    There is now a further tool here: https://tools.wmflabs.org/scholia/

    Veronica Santos said:
    May 27, 2021 at 3:32 am

    Which Wikidata schemas (from this list -> https://www.wikidata.org/wiki/Wikidata:Database_reports/EntitySchema_directory) does Scholia use?

      Finn Årup Nielsen responded:
      May 28, 2021 at 2:23 pm

      Scholia does not currently use the Entity Schema of Wikidata. The way that the queries are setup is by looking at how items are conventionally used. Sometimes items are used in a (slightly) different way and we may not catch these variations. For scholarly documents, discussion of which properties should be used are often done under the “WikiCite” umbrella.

Leave a comment