Rate article/subject importance.
You give it an article and it tells you how important that article probably is to wikipedia, generally.
(optional) You give it an article and a WikiProject and it tells you how important it thinks that article is to the specified WikiProject.
Wiki thing it helps with:
- SuggestBot could better help editors find the most impactful things to work on.
- We could take better measurements of productivity in Wikipedia: https://meta.wikimedia.org/wiki/Research:Measuring_value-added
- We can sort lists of work on Wikipedia by importance
- We might be able to recommend articles to WikiProjects
- Things that might helps us get this AI built (optional): **
- Domain-specific datasets about importance (e.g. list of biographies)
- If a way is found to automatically extract relevant "keywords" or topics from a Wikiproject, a similarity measure between those and the article could be a reasonable metric for importance of article to that Wikiproject.
- Maybe something like topic modelling on a Wikiproject for above ^ as an idea.
- Wikipedians label importance: https://en.wikipedia.org/wiki/Wikipedia:Version_1.0_Editorial_Team/Release_Version_Criteria#Importance_of_topic
- Views and links seem like good proxies for article importance: https://meta.wikimedia.org/wiki/Research:Measuring_article_importance
- You can get pageview rates from an API: https://wikitech.wikimedia.org/wiki/Analytics/PageviewAPI
- You can get links from tha pagelinks table: https://www.mediawiki.org/wiki/Manual:Pagelinks_table
- You and query tables with Quarry: https://quarry.wmflabs.org/
- Wikiprojects label importance independently. E.g https://en.wikipedia.org/wiki/Waffle is Top-importance for WikiProject Breakfast, but only High-importance for WikiProject food and drink.
- Maybe we can use within-WikiProject link stats and within WikiProject view rates to model that more accurately.
Research Project page: