Title: Hyperbolic Knowledge Graph Embeddings for User Interests
Speaker: Federico López (HITS)
Abstract
Generating recommendations with a knowledge graph as side information has attracted considerable interest in recent years. Such an approach can not only alleviate data sparsity and cold start issues for a more precise recommendation but also provide explanations for recommended items. In this talk, we propose a knowledge graph-based recommender system that operates in hyperbolic space. The model combines expressive operators to exploit multiple relations between the items. Moreover, it learns personalized preferences to provide accurate and explainable recommendations for each user. We present results on two datasets and different by-products derived from the application of hyperbolic geometry.
