The on-wiki version of this update can be found here:
https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Updates/2022-10-27
--
Thank you, fellows!
In April we shared with great happiness that
Google.org supported the
development of Abstract Wikipedia and Wikifunctions with a generous
fellowship program. Six fellows joined us in April
<https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Updates/2022-04-12> and
May for a period of six months, and as their fellowship has come to an end,
it is time to take a look back and appreciate their achievements. It was a
pleasure to work with the fellows, and some of us had the lucky opportunity
to meet
<https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Updates/2022-10-05>.
*Olivia Zhang* joined the fellowship as a product manager, and she filled
that role for the performance workstream and beyond, organizing and
prioritizing the numerous requirements, and outlining a product narrative
for performance
<https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Wikifunctions_performance_product_narrative>
at
Wikifunctions. She also created and managed an overall launch document for
Wikifunctions, allowing us to get a better view of our status and the work
ahead. The launch document will now be taken over by our product manager
Rebecca
<https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Updates/2022-07-20>,
who recently joined the team. Olivia helped us understand and answer
important questions about Wikifunctions's ultimate scope and how to best
achieve it.
Thanks to *Mary Yang* and her work we could safely launch the Wikifunctions
Beta <https://wikifunctions.beta.wmflabs.org/wiki/Wikifunctions:Main_Page> in
time for Wikimania. She set up the availability and performance monitoring
infrastructure not only for the wiki itself, but also for the supporting
services such as the evaluator
<https://gerrit.wikimedia.org/g/mediawiki/services/function-evaluator> and
orchestrator
<https://gerrit.wikimedia.org/g/mediawiki/services/function-orchestrator>.
Her work is crucial in preparation for the general launch of Wikifunctions.
She contributed numerous patches to the performance workstream. This also
allowed Mary to dive and understand our architecture deeply, and she
provided us with insights into the strengths and weaknesses of our
architecture.
*Eunice Moon* supported the fellowship as a program manager. She supported
the fellows in their work, particularly with the implementation of the
natural language workstream, involving external members from the community
and academia. Together with our community relation specialists and other
folks in both organizations, she set up and planned out a series of posts
on Diff
<https://diff.wikimedia.org/2022/09/21/the-state-of-abstract-wikipedia-natural-language-generation/>
further
outlining the project and its aims. She also was the main organizer behind
the in-person Zürich Fellowship offsite which the fellows attended.
With *Ori Livneh*, we were able to welcome back an old friend. Ori was not
only the tech lead for the fellowship program, but also was an invaluable
translator between the world, the culture, and the language of Google and
Wikimedia, allowing for the other fellows to work smoothly together. Ori
was a catalyst for the launch of the Wikifunctions Beta
<https://wikifunctions.beta.wmflabs.org/wiki/Wikifunctions:Main_Page> in
time for Wikimania. He also led the work on a thorough evaluation of the
strengths and weaknesses of the architecture of Abstract Wikipedia and
Wikifunctions, which is going to be published soon.
*Ariel Gutman* brought his many years of experience in natural language
generation to the project. Together with community members such as Mahir
Morshed, external experts such as Maria Keet, and other Googlers such as
Kutz Arrieta, they started the natural language generation workstream.
Ariel boldly sketched out a proposal for the natural language generation
architecture
<https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Updates/2022-05-27> for
Abstract Wikipedia, outlined (together with Maria Keet) a template language
<https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Template_Language_for_Wikifunctions>
to
be used, and discussed the progress in a recent Diff post
<https://diff.wikimedia.org/2022/09/21/the-state-of-abstract-wikipedia-natural-language-generation/>.
He also created a Lua-based prototype
<https://meta.wikimedia.org/wiki/Module:Sandbox/AbstractWikipedia>, in line
with this architecture, demonstrating that the proposal is feasible; he
will soon present this work in detail.
When we learned that *Ali Assaf* had written his PhD on lambda calculus, we
were very excited. The lambda
<https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Updates/2022-02-04> in
the name "WikiLambda
<https://www.mediawiki.org/wiki/Extension:WikiLambda>"
pays homage to the inspiring role of lambda calculus for Wikifunctions. Ali
dived right into formalizing the underlying function model, putting the
whole project on a stronger conceptual and practical foundation. He
discovered and fixed plenty of bugs in the orchestrator
<https://gerrit.wikimedia.org/g/mediawiki/services/function-orchestrator> where
composition evaluation is implemented. Ali’s main result is the semantics
of Wikifunctions
<https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Semantics_of_Wikifunctions>
documentation
and further essays on generic function types
<https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Generic_function_type>,
local
vs. global keys
<https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Local_vs_global_keys>,
and the semantics of validation
<https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Semantics_of_validation_in_Wikifunctions>
.
To all the fellows: thank you!
Three more fellows
<https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Updates/2022-07-20> remain
with the team, and will stay with us until the end of the calendar year,
Dani de Waal, Sandy Woodruff, and Edmund Wright.
The
Google.org fellowship program is designed to let Google employees work
pro bono with nonprofits and civic entities. Fellows build open-source
solutions to equip their hosting organizations with better capabilities to
reach their goals. As you can see with the contributions of the fellows to
Abstract Wikipedia and Wikifunctions, this is an opportunity to benefit
from people with unique talents, and to allow them to do pro bono work. We
want to thank the fellows for their willingness and interest to work with
us, and we hope that they had a great time. We also want to extend our
gratitude to
Google.org for allowing us to participate in this programme.
Development update for the week of October 21, 2022
Experience:
- Made type expansion unnecessary in the evaluator
- Fixed more FE bugs
- Completed proposal for front-end error handling
<https://etherpad.wikimedia.org/p/FE_error_management_proposal>
- Merged ZObjectDiffer
Meta-data:
- Drop back-compat. code in PHP and Vue layers (T291136
<https://phabricator.wikimedia.org/T291136>)
- Developed proposal for readable summaries of all error types (T312611
<https://phabricator.wikimedia.org/T312611>)
Natural Language Generation:
- Categorized discussion points on Abstract Representation
- Refined the template language
- Patch
<https://gerrit.wikimedia.org/r/c/mediawiki/extensions/WikibaseLexeme/%2B/828887>
of
Ndebele language codes got pushed to Wikidata