Research talk:Recommending links to increase visibility of articles

Feedback from Wikimania 2024

Context: I presented at Wikimania 2024 and that presentation included the learnings from orphan articles research. That presentation inspired multiple follow-up conversations with the community. I'm capturing the main pieces I have noted down below.

1. I understand from this research you know that some languages do really well in de-orphanizing articles (e.g., nl) and some need extensive work (e.g., kk). Multichill mentioned that in Dutch Wikipedia they have/had someone on the task for de-orphanization. If ever qualitative studies are needed to understand "how" to better support editors for de-orphanization, I suggest considering talking with this person or other people who may be in similar roles within the community.
2. Nemo_bis brought up the challenge of de-orphanization particularly if the anchor text does not exist in another article. We had some conversations about this and he had some ideas about how to simplify the task or at least identify subsets of the articles that may be easier to de-orphanize. This can be particularly relevant if the task is targeted to newcomers. Some qualitative work (if not already done) can help understand this space better.
3. If Growth team decides to prioritize de-orphanization for link recommendation, I can say that the sentiment of the conversations were very positive about it. Many also resonated with the fact that we may want to prioritize de-orphanizing biographies of women given the structural bias that is currently in place.
4. Many took note of the point around the structural biases that are in place unless we address the orphan article issue. I trust that they will take action on it (and I explicitly encouraged people to not wait for WMF and that all of us have a role and we can take action on this insight. For example, Claudia Serbanuta who presented about Wikimartisor shared interest to help address this issue for biographies of women.
5. I think the case for de-orphanization is quite clear with the results of this research. However, if we have bandwidth to work on implementing a metric around it (task), it can certainly help mobilize more folks. cc Miriam and Xiao.

Thank you, Martin, Akhil and Bob, for this research and the ongoing work. There is a lot of support and enthusiasm for addressing the issues identified through this work and it will be really great if we see some implementations through products for addressing it. LZia (WMF) (talk) 08:25, 12 August 2024 (UTC)