Hi James,
Thanks for your message. I must say that we had a moment of stepping back and enjoying to read this email, so thanks for that! :) It's always good to know people are out there to support. Sometimes our world can get really lonely. :)
A few things about the methodology in the paper and where we are with it today: * As you may know, the recommendation APIs related to that research can be reviewed at: https://www.mediawiki.org/wiki/GapFinder/Developers . They all go under the umbrella of GapFinder as a tool/project.
* None of the recommendations in the link above are as sophisticated as the one tested for the study in the paper. There are historical reasons for that, but looking forward: ** We have set a goal for Apr-June 2018 to improve the recommendation API and bring it to as close as possible to the paper. We have called this goal out [1]. We're hoping that by the end of June 2018, the API is in a much better place. ** The translation API linked in GapFinder page is surfaced in Content Translation today, on top of recommend.wmflabs.org, which is already great. I hear from Content Translation team that 20% of their created articles come from the recommendation API. I'd love to see us not stop here. The nice thing about having an API is that more tools can be built on top of it. ** We would love to work with the community to make the API more useful for places such as https://en.wikipedia.org/wiki/Wikipedia:Community_portal on English Wikipedia. This kind of place seem to be natural place for doing more collaboration in this space.
Given that you've read the paper and you know what's in it, if you see an opportunity for bringing it to the attention of specific projects, please do so. We do the same on our end. We're trying to avoid what I call raw advocacy which can put a disproportionate pressure on developers and product owners to use it. :) Only where you see reasonable; and we will be of course, thankful! :)
Best, Leila
[1] https://www.mediawiki.org/wiki/Wikimedia_Technology/Goals/2017-18_Q4#Program... (Outcome 2, Objective 1)
On Mon, Mar 26, 2018 at 11:02 PM, James Salsman jsalsman@gmail.com wrote:
Leila, since https://arxiv.org/pdf/1604.03235.pdf worked so well, is it going to be implemented widely, or is it shelved? Does it need advocates?
Wiki-research-l mailing list Wiki-research-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wiki-research-l