KNNs of Semantic Encodings for Rating Prediction
This is an incremental improvement for recommendation systems, enabling review-based explanations.
The paper tackles rating prediction by representing user preferences as a graph of textual snippets from reviews, using semantic similarity for edges, and shows that this approach outperforms strong collaborative filtering baselines.
This paper explores a novel application of textual semantic similarity to user-preference representation for rating prediction. The approach represents a user's preferences as a graph of textual snippets from review text, where the edges are defined by semantic similarity. This textual, memory-based approach to rating prediction enables review-based explanations for recommendations. The method is evaluated quantitatively, highlighting that leveraging text in this way outperforms both strong memory-based and model-based collaborative filtering baselines.