search

WDscholia / scholia

Wikidata-based scholarly profiles
https://scholia.toolforge.org
Other
215 stars 78 forks source link

Feedback on the topic profile for SARS-CoV-2 #1106

Open Daniel-Mietchen opened 4 years ago

Daniel-Mietchen commented 4 years ago

At the Biohackathon, I noticed @bmeldal offering advice on UX issues, and since we have a good number of these, I replied to her by inviting feedback on the topic profile for SARS-CoV-2.

Here is what she came back with, interspersed with some quick responses:

Tables look fine. "Publications per Year" Doesn't really make much sense for a single year. Maybe per month?

  • The tool is not specific for COVID-19 but can visualize live data held and community-curated in Wikidata about things like topics, people, institutions, proteins, events and such. Besides presenting the information for research and education purposes, it thus also assists data curation by helping to identify gaps and inconsistencies, and by facilitating these to be addressed. For topics, our default is to display publications by year, but each query is editable, so users can adapt it to things like months, day of the week or decades. I'm not sure the orcid search string is very helpful when you can't disambiguate it. Maybe just provide a message where the query fails?
  • ORCID search: we included that in order to encourage users to help curate the content of the profiles. We also have /missing pages for such purposes, e.g. https://tools.wmflabs.org/scholia/topic/Q82069695/missing , which also holds part of the answer to your question about China (below) - many author names have just not been mapped to identified authors yet, and this is especially true for Chinese names. Author graph: Zooms in when you scroll through it - I know, it's a know issue with embedded widget, we have that, too :disappointed:
  • What's your tool? Author graph: Nodes can be moved but don't stay in new position. Some nodes overlap nearly 100% so it's difficult to read. E.g. Christian Drosten and Stanley Perlman - but now they stay apart! Seem to behave a bit differently depending on how far one zooms in.
  • Yes, would be nice to have that fixed, but it's upstream for us, so not straightforward. Co-occurring map & author score bubble: I don't actually understand the use case of these. Sorry.
  • This could for instance be journalists or educators trying to identify topic-related research near them, or a local expert on the matter. The bubble is more useful once the citation graph is more complete - see the Zika virus for comparison: https://tools.wmflabs.org/scholia/topic/Q202864 . A very generic use case for maps is to help curate geo data, e.g. things like railways in the middle of the sea because of lat/ long hickups. I just discovered the side bar feature of changing the visualisation type. For those examples I looked at the chosen one looks best.
  • Thanks for checking that too. Citation table is interesting. Can it be split by preprints and peer-reviewed? Just because we are looking at such a short time span, it might be interesting to see how many papers have already gone through full peer review and are cited.
  • That's tricky but together with @konrad and others, we are looking into making some version of this possible. Challenge is that few sources indicate details about the peer review in a way that is friendly to machines or even crowds. But preprints posted on preprint servers can be identified and displayed as such. Map of organizations associated with works about the topic: That's an interesting and not unexpected distribution. Although, I thought there was more in China.
  • This depends on annotation of papers with topic and authors, authors with institutions and institutions with geolocations. Many of these are missing or incomplete right now, and in a way, Scholia helps identify gaps in coverage, which can then help prioritize Wikidata curation efforts. Maybe the dots can be proportional to the number of publications that came from that organisation? Or proportional to no. of authors? Lots to play with here :wink:
  • Yes, dots indicate number of publications associated with the locality, but agree the scale is not ideal. A patch has been committed to improve that situation (see screenshot). Screen Shot 2020-04-09 at 21 40 02
bmeldal commented 4 years ago

Author graph: Zooms in when you scroll through it - I know, it's a know issue with embedded widget, we have that, too 😞 What's your tool?

https://www.ebi.ac.uk/complexportal/home E.g. Reactome widget in example https://www.ebi.ac.uk/complexportal/complex/CPX-2158 Actually, Reactome fixed that (I forgot they did!) but originally the curser took over control of the widget and you had to click out to scroll further down. Their developer has left so you'd have to contact help@reactome.org to find out what they did to stop this.

Citation table is interesting. Can it be split by preprints and peer-reviewed? Just because we are looking at such a short time span, it might be interesting to see how many papers have already gone through full peer review and are cited. That's tricky but together with @konrad and others, we are looking into making some version of this possible. Challenge is that few sources indicate details about the peer review in a way that is friendly to machines or even crowds. But preprints posted on preprint servers can be identified and displayed as such.

I have seen Biorxiv entries that have a link to the peer-reviewed publication. No idea how that looks on the API and I'm sure you looked at that. Also, EuropePMC pull in preprints and give them PPRxxxxx IDs. We haven't yet figured out how they link PPR IDs and PMID later on but there must be a way. You probably investigated this all already, though!