What is it about?
This work shows how the usage of First-Order Logic Rules in the Graph Embedding process can improve the quality and the precision of the generated embedding, and in turn improve the quality of a neural recommender system fed with those embeddings; this kind of embedding can be said to be a neuro-symbolic technique.
Featured Image
Photo by Sigmund on Unsplash
Why is it important?
The key contribute of this work is adapting this kind of neuro-symbolic graph embedding technique in a recommendation scenario; to best of our knowledge, this technique was never investigated in this area. Results show how adding logic rules not improves the quality of the recommendation, and also how this strategy can outperform several competitive knowledge-aware baselines.
Perspectives
I hope this work represents the first step in the study of this kind of techniques and application in this area; writing this paper was satisfying and pleasant thanks to my co-authors who always supported me in this research.
Giuseppe Spillo
Universita degli Studi di Bari Aldo Moro
Read the Original
This page is a summary of: Knowledge-aware Recommendations Based on Neuro-Symbolic Graph Embeddings and First-Order Logical Rules, September 2022, ACM (Association for Computing Machinery),
DOI: 10.1145/3523227.3551484.
You can read the full text:
Contributors
The following have contributed to this page
