A Healthy Food Recommender System Using Collaborative Filtering and Transformers

Mustafa Zaki, Junyuan Lin, Mandy Barrett Korpusik

Keywords: Collaborative Filtering, EASE, Nutrition, BERT

Abstract:

Unhealthy eating habits are a major contributing factor to public health problems such as the globally rising obesity rate. One way to help solve this problem is by creating systems that can suggest better food choices in order to improve the way people eat. A critical challenge with these systems is making sure they offer 1) suggestions that match what users like, while also 2) recommending healthy foods. In this paper, we introduce a novel food recommender system that provides healthy food recommendations similar to what the user has previously eaten. We used collaborative filtering to generate recommendations and re-ranked the recommendations using a novel health score and a BERT embedding similarity score. We evaluated our system on human subjects by conducting A/B testing on several methods deployed in a web application.

Primary Area: other topics in machine learning (i.e., none of the above)

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 5201