Proposal on Modeling Personal Values from Review Text Using Large Language Models

Abstract

This paper proposes a method for constructing personal value-based user models from review texts using LLM (Large Language Models). The RMrate (Rating Matching Rate) has been proposed as a metric to quantitatively assess the intensity of user preferences towards item attributes when selecting items, and has been applied to personal value-based models. Although its effectiveness in information recommendation has been demonstrated, existing methods require explicit attribute evaluations. To address this issue, the proposed method calculates RMrate by applying LLM to extract the evaluation polarity of item attributes mentioned in reviews through prompting. This paper conducts experiments with movies as the target items, demonstrating the accuracy of extracting evaluation polarities and the effectiveness of the proposed method when applied to a recommendation system.