This week's update can be found on-wiki here:
https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Updates/2021-03-25
--
Back in December 2020, we introduced the Abstract Wikipedia list to Aisha
and Jade, Outreachy interns participating in Abstract Wikipedia /
Wikifunctions data science.[1]
We want to share the exciting results of their work. During the past few
months, Aisha and Jade developed a Vue.js web application to help people
find user generated code across the MediaWiki Scribunto modules written in
Lua. Their creative approach involved a complex data pipeline using
Wikimedia Cloud Services which analyzed the usage and similarity of code
across different language editions of Wikipedia and its sibling projects.
This approach is intended to aid users as they consider the problem space
of consolidation of certain types of code into pure, re-usable functions in
the Wikifunctions project.
The application can be found here:
https://abstract-wiki-ds.toolforge.org
We hope you enjoy the following 3 minute demo video of the web-based tool:
https://www.youtube.com/watch?v=M31scvrfGdE
Here is the longer 41 minute audio discussion with Aisha and Jade and the
rest of the Abstract Wikipedia team about their data science work:
https://commons.wikimedia.org/w/index.php?title=File%3AAbstract_Wikipedia_D…
You can review the extensive data analysis and pipeline development and
web-based tool material, including Jupyter notebooks (on the PAWS public
notebook system) and code hosted in Wikimedia GitHub, starting from here:
https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Data#Get_important_modul…
We want to thank Aisha and Jade for their excellent work during the
internship, and we hope they return in the future.
If you'd like to collaborate on the data pipeline, further analysis, or the
web-based tool, please do drop a line on the abstract-wikipedia mailing
list and we'll see if we can connect on Phabricator and GitHub to grow this
work further. We believe the approach taken here, if further generalized,
may also be of interest as we look deeper at the natural language
constructs across Wikipedias and sibling projects for Abstract Wikipedia,
and as more folks explore topics like global templates as well.
We'd like to thank Srishti and all of the Wikimedia Foundation's Developer
Advocacy unit for their support in Outreachy efforts and Wikimedia Cloud
Services capabilities, as well as for the encouragement from Isaac in
Wikimedia Foundation's Research unit to foster this collaboration with both
Aisha and Jade - they formed a productive partnership and as the mentoring
group we were delighted to work with Aisha and Jade.
We're delighted to have such a supportive global movement and knowledge
network helping to increase the impact and awareness of this valuable
initiative.
-The Abstract Wikipedia Team
https://meta.wikimedia.org/wiki/Abstract_Wikipedia
P.S.: We also wanted to let you know that an overview of how all the pieces
fit together is now published in the April 2021 edition of the
Communications of the ACM:
https://cacm.acm.org/magazines/2021/4/251343-building-a-multilingual-wikipe…
[1]
https://lists.wikimedia.org/pipermail/abstract-wikipedia/2020-December/0003…