Scribe/Monthly Updates
< Scribe
Home | Blog Posts & Presentations | Get Involved |
Scribe
Helping editors of under-resourced languages to create new high-quality Wikipedia articles |
Resources
|
Updates 2021
edit- The paper "References in Wikipedia: The Editors’ Perspective" was accepted at WikiWorkshop 2021
- Presentation of Scribe at the Qurator Conference 2021 description of the presentation
Updates 2020
edit- Scribe was deployed on Arabic and Catalan Wikipedia
Monthly Report June/July
edit- Addition of the new scores to the database, change the database schema, add tables
- We are working on the last bug fixes for the interface
- We are changing the interface to include the credibility scores
- We contacted Catalan Wikipedia to deploy the Scribe gadget on their Wikipedia
Monthly Report May/June
edit- We implemented the filtering mechanisms for references. After discussions, we decided to display all scores that we extract rather than just not displaying the reference at all. This leads to some changes to the interface we are still discussing.
- The new content needs to be added to the database. Additionally, new tables have to be created for the reference meta-information and we are creating lists of red links for the articles that can be created with Scribe.
- Eugene created tables to record statistics about the Scribe usage.
- After talking to Arabic community members, one of the admins deployed Scribe as a gadget on Arabic Wikipedia. \o/
Monthly Report April/May
edit- We focused on how to filter references after retrieving them from a search engine for this we set the following first steps:
- The reference has been used on the Wikipedia overall and in particular in this domain (domain retrieved from Wikidata)
- The reference is the official website of a Wikidata entity and then we can filter on properties of the Wikidata entity (e.g. is the entity an **instance of** a news agency or a social network etc.)
- For the above factors, we planned and implemented the following:
- Get a cloud server to run experiments with large datasets
- Create whitelist for all of the Arabic and Catalan Wikipedia for filtering
- Start the coding for the match with Wikidata and extract information
- In terms of the interface, we redesigned the interface after the conversation with the WMF UX team
- The interface now has elements embedded directly in the visual editor rather than as a layover
- This has been already implemented in the new gadget
- Demo video to follow
- We wrote mails to the Arabic and Catalan mailinglist to get the gadget deployed on their respective Wikipedias, but have not heard back yet
Monthly Report March/April
edit- Integrating Scribe as a Module in the Mobile extension
- Meeting with the WMF UX team recording meeting notes
- Integrating new with recommended design flows from the meeting with the WMF UX team
- Research on reference extraction, collections references for 200 Ar and 200 Ca Wikidata Items.
Monthly Report February/March
edit- New prototype for Scribe on test.wikipedia.org
- Participation in design review of WMF to discuss Scribe interface (desktop and mobile)
- Redesign of mobile logic
- Research on reference extraction, first implementations
- Plan community outreach
Monthly Report January/February
edit- Become interface admin on test.wikipedia.org to install gadgets
- Deployment of the gadget for desktop on test.wikipedia.org
- Bugfixes and others
- Different plans for reference extraction
Monthly Report December/January
edit- Code for whitelist for references most used on each Wikipedia language: [1]
- Presentation of Scribe at Credibility Coalition & NewsQ Workshop in New York, USA
- Presentation of Scribe at Citizen Beta, London, UK
- Organization of meetup for researcher and Wikimedians to exchange in London, UK
Monthly Report November/December
edit- Working on creating a prototype for Scibe
- Eugene created a first version of Scribe as a gadget
- Eugene created a tool for the backend of Scribe on toolabs
- Selecting examples for the test version of GLAM institutions in Arabic and Catalan
- Adding examples to the database
- Automated section selection for GLAM institutions, creating templates of sections for different types
- Exchange with organizers of WikiCredCon about references on Wikipedia
- Created a project on github for the code (both for the interface and the research part)
Monthly Report October/November
edit- Hired a developer (Eugene)
- Participation at WikidataCon
- Presented a poster about Scribe
- Presentation on Wikidata's future with regards to supporting underrepresented language communities on Wikipedia, including Scribe
- Made connections with different language communities interested in testing Scribe
- Kick-off development meeting, setting up development infrastructure
- Spread the word about Scribe at WikiIndaba 2019 and gathering ideas from participants on how to make the tool better
Monthly Report September/October
edit- Recruitment of a developer
- To get a more diverse set of applicants, we extended the time looking for a developer and are interviewing two more applicants
- Participation in WikiArabia 2019
- To build stronger connections with the Arabic community, we participated in WikiArabia 2019
- We gave two presentations: A look into the future: Supporting Arabic Wikipedia with Wikidata and ArticlePlaceholder and Research and Arabic Wikipedia: Finding innovative ways to support editors
- We made connections with Arabic Wikipedia editors to test the prototype once it’s up
- We are working on the minimal technical requirements
- Laying out what is the requirements of the first prototype
Monthly Report August/September
edit- Wikimania 2019 Hackathon:
- Alongside with Joe Reeve, we made a mobile web early prototype of Scribe that works on the Wikipedia visual editor. The demo suggests chunk editing spaces to new editors based on sections and references from Wikidata and the web. This demo works on the Czech language.
- Participation in Wikimania 2019 with two presentations:
- In the growth track: https://commons.wikimedia.org/wiki/File:Scribe_Presentation_Wikimania_2019.pdf
- In the Research track: https://commons.wikimedia.org/wiki/File:How_can_nlp_research_help_(1).pdf
- Published blogpost about the project kickoff: https://medium.com/scribewikimedia/scribe-our-project-is-kicking-off-8b3eb5c830c9
- Writing and disseminating the hiring position https://meta.wikimedia.org/wiki/Scribe/Job
- Scheduling interviews with the applicants
- Exchanging with WMF about the project (new readers team, growth team)
Monthly Report July/August
editThe first month of our project focused on administration work: We had to figure out how we would tax the project funds, how to hire and pay a developer and many small tasks, that come with the start of the project.
At the same time, we wanted to prepare the outreach and possible collaborations with the communities, that we believe could be interested in our work:
- For the research community, we plan to find researchers that are interested in collaborations in the topic around Scribe
- We wrote a blogpost to start the project and to introduce which areas we plan to include
- We were asked to submit a video to the conference Worlds of Wikimedia
- For the Wikimedia community, we want to introduce the editors to the project, collect their opinion and understand how they currently edit to draw from their experience
- We started submitted and prepared a presentation at Wikimania, as part of the community growth space
- We got an ethics approval for interviews from the University of Southampton to conduct a series of interviews with editors
- We started a survey and interviews with editors across different language Wikipedias that give us first insights into their editing. We plan to continue those interviews, the link to sign up for participation can be found here
- Finally, one of the communities we want to address is the technical volunteer community of Wikimedia Projects. We believe to build a sustainable project, we need to include the technical volunteers from the very beginning of the project
- We participated in the Wikimedia hackathon (a blogpost about the experience can be found here) with a poster
- We created a Scribe Phabricator Board to keep track of tasks in a way that makes it easy for interested community members to participate
- We worked with two programmers on the prototype of Scribe at the hackathon. We are happy to announce that there is a very first prototype in the form of a user script, that gives an idea of how the project works
- We are participating in the hackathon leading to Wikimania 2019 in Sweden. Again, we plan to show volunteers the different possibilities to work with us on Scribe's technical part