Machine Learning
Community Rules:
- Be nice. No offensive behavior, insults or attacks: we encourage a diverse community in which members feel safe and have a voice.
- Make your post clear and comprehensive: posts that lack insight or effort will be removed. (ex: questions which are easily googled)
- Beginner or career related questions go elsewhere. This community is focused in discussion of research and new projects that advance the state-of-the-art.
- Limit self-promotion. Comments and posts should be first and foremost about topics of interest to ML observers and practitioners. Limited self-promotion is tolerated, but the sub is not here as merely a source for free advertisement. Such posts will be removed at the discretion of the mods.
This is kind of a low effort post but I’ll bite. I think your issue, if you’re only now building a platform, is lack of data. Recommender systems typically require data, be it history of the users’ interactions or at least data about the items being recommended. If you are creating a platform entirely from cold start this will be a challenge. If your items have semantic meaning, maybe generate text embeddings for the names/descriptions and then on the product page add a tab such as “Similar items…” which retrieves the top 5 neighbors per those text embeddings. Boom, recommender system.
Personalized recommender system? No, not doable. This requires separate setup, model serving, regular retraining since typical collaborative filtering does not support adding users and items. Besides, it makes no sense if you don't have prior data.
However, implementing popularity-based system, without personalization, is possible. This is typically just simple Bayesian statistics, doable in 10-15 lines of Python. See e.g.:
- https://arpitbhayani.me/blogs/bayesian-average/
- https://www.evanmiller.org/how-not-to-sort-by-average-rating.html
- https://www.evanmiller.org/ranking-items-with-star-ratings.html