Diego generated section mappings (Gdrive). We want to make these mappings available via a new section recommendation API.
A/C
- Get MySQL storage for data
- Import data into MySQL
- Create an API. The code will be part of recommendation-api.
Description
Description
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | diego | T203046 Output 1.4: Public test APIs corresponding to section recommendation algorithms | |||
| Resolved | None | T210530 Expose section mappings via an API |
Event Timeline
Comment Actions
@leila is the task description what you had in mind when you were talking about creating an API for the paper?
Comment Actions
@bmansurov almost. I'll open a thread about this with you, Bob, and Diego now to find out the exact format of the data and output of the API
Comment Actions
@Pginer-WMF We will soon have a working model that aligns article sections across some/many languages. We are building an API to surface the alignment, and we'd like to understand what kind of input/output would be interesting to Content Translation (even if it's not immediately used by the tool). The data that we have, for example, can look like:
from_language, section_from, to_language, section_to, probability
Comment Actions
The most immediate use would be similar to a translation service. If I'm translating an article from English into Russian, when I add the "In popular culture" section from the original article, we may want to check that "В культуре" is the equivalent in Russian and add that. That would be especially useful for languages where machine translation services are currently not available.
Comment Actions
This has been solved here: https://meta.wikimedia.org/wiki/Research:Expanding_Wikipedia_articles_across_languages/Inter_language_approach#Results
Check an example here: https://secrec.wmflabs.org/API/alignment/en/ja/Work