Translational models for item recommendation

Translational models have proven to be accurate and efficient at learning entity and relation representations from knowledge graphs for machine learning tasks such as knowledge graph completion. In the past years, knowledge graphs have shown to be beneficial for recommender systems, efficiently addressing paramount issues such as new items and data sparsity. In this paper, we show that the item recommendation problem can be seen as a specific case of knowledge graph completion problem, where the "feedback" property, which connects users to items that they like, has to be predicted. We empirically compare a set of state-of-the-art knowledge graph embeddings algorithms on the task of item recommendation on the Movielens 1M and on the LibraryThing dataset. The results show that translational models outperform typical baseline approaches based on collaborative filtering and popularity and that the dimension of the embedding vector influences the accuracy of the recommendations.