Wikipedia Science Conference/Submissions/Hackathon
When: Saturday 5th September
Where: Basement, Development House (Wikimedia UK HQ)
- Wikimedia UK
- 4th Floor, Development House,
- 56-64 Leonard Street
- London
- England
- EC2A 4LT
- Download address as vCard for your address book.
- Coordinates
- 51.524483, -0.084477
Wikimedia UK's Friendly space policy will apply at this event.
This is accepted and will go ahead with the cooperation of Wikimedia UK. MartinPoulter (talk) 13:18, 9 May 2015 (BST)
Personal details
- Your name
(talk) 13:07, 1 September 2015 (BST)
- co-organizers most welcome
- Affiliation
- How can we contact you?
- daniel.mietchen
okfn.org
- Where will you be travelling from to attend this conference?
- Probably from Germany
- Availability
- I shall be there for both days and the hackathon.
Session details
- Conference themes
Which conference themes does your session address? Delete as appropriate. Follow the link for more details and links.
- Wikipedia and/or Wikimedia as platforms for promoting informed public discussion
- Wikipedia and/or Wikimedia as platforms for research (including citizen science)
- Wikipedia and/or Wikimedia as models for scientific publishing
- Wikipedia and/or Wikimedia as platforms for scientific education
- Under-represented groups in STEM subjects
- Type of session
- Satellite event on Saturday September 5th
- Further details
- Location
- We have reserved two rooms in the basement area of Development House (the building housing WMUK in East London) for this day
- Title: Wikimedia & Research hackathon
- Abstract: There are many technical aspects to the interaction between the Wikimedia and research communities and platforms. This day is dedicated to exploring them together with others who share an interest in these matters. This may include work on
- or many other things. Suggestions are most welcome.
- Note
- There will be a satellite hackathon at EBI in Hinxton on September 4. Details here.
Hackathon Summary
Present:
Discussions
1. Classification task force/notice board for Wikidata
It was pointed out that a wide range of different decisions are being made by Wikidata editors when trying to classify items using the "Instance of" and "subclass of" properties, and that the structure that is defined by these properties is really crucial for running efficient and successful queries on the data. There is a risk of getting into a similar mess as the Wikipedia category system if there isn’t some careful consideration keeping it as uniform as possible. All present agreed that having a task force, or noticeboard would be a great help. Basically a place where you can always post your questions about classifying a particular item (or entire class of items). A core group of community members with experience in classification could then answer those queries and help encourage more uniform usage.
2. Tool for determining why a particular item is a result of a query
The tool would take in 2 bits of information:
- WDQ query
- Q number of an item that should not have been in the list of results to the query given.
It would then display the parts of the class tree that connected the item to the query. The idea being that you can quickly identify and fix the exact item that has an incorrect class statement on it.
For example, the tool could display this tree if you were trying to find out why a psychopath turned up in your query for people with occupation = scientists (or any subclass of):
- Scientists
- Physicist
- Psychopath
- Physicist
After-hackathon-thought: Although this be very handy as a stand alone tool, an option in AutoList would also be great! So you could run a query, then click on an item to show why it is in the list of results for the query.
3. Ranking system for Wikidata items
Histropedia is planning on creating a ranking system for Wikidata items, based on Wikipedia backlink counting across all languages. It seems evident that this will be useful to many more projects than just Histropedia. Obvious uses spring to mind, like having AutoList results ranked by ‘importance’, sorting Wikidata search results, or selecting the ‘top 50’ results of a query. If lots of people feel this will be useful to have, we should consider building and maintaining the rank list on Labs instead of developing it externally.
4. Tool for plotting a timeline of references using Histropedia
A tool was proposed that would allow you to see the references for any Wikipedia article visualised on a Histropedia timeline (by publication date). The tool would retrieve the list of references from a Wikipedia article, then use the Wikidata items for each reference to pull the necessary data for plotting a timeline (e.g. date of publication, image and link to the reference on Wikisource). Obviously the usefulness of the tool will depend on the number of references that we can get properly described on Wikidata, so hopefully it could serve as good motivation for editors.
Follow up tasks
- Post a message on Wikidata project chat about the Classification task force / notice board idea.
- The notice board now exists at Wikidata:Classification noticeboard.
- Plan and propose tool for identifying –why-- items incorrectly appear in a query (run this past Magnus for feasibiliy check)
- Post message to project chat about using Wikipedia backlink counting as a ranking system for Wikidata items to gauge interest.