KNNs of Semantic Encodings for Rating Prediction

TL;DR

This paper introduces a novel method for rating prediction that uses semantic similarity of review texts to represent user preferences, enabling more explainable recommendations and outperforming traditional methods.

Contribution

It proposes a new textual, memory-based approach using semantic similarity graphs for rating prediction, providing review-based explanations and improved accuracy.

Findings

Outperforms traditional collaborative filtering methods

Enables review-based explanations for recommendations

Demonstrates effectiveness through quantitative evaluation

Abstract

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.