Collaborative Filtering-based Movie Recommendation Services Using Opinion Mining

Abstract

This article explores the fusion of collaborative filtering (CF) techniques with opinion mining methodologies to enhance movie recommendation systems. The proposed approach harnesses user-item interaction data to employ CF algorithms for generating initial movie recommendations. Furthermore, opinion mining techniques are integrated to analyze textual reviews and extract implicit sentiment signals associated with movies. Leveraging sentiment-aware adjustments, the model refines CF-based recommendations by incorporating user sentiments expressed in reviews. In specifically, conducting aspect-based sentiment analysis on movie reviews leveraging a Long Short-Term Memory (LSTM) neural network architecture. The methodology involves applying LSTM-based sentiment analysis to extract sentiments from movie reviews, followed by aspect identification and sentiment assignment to specific movie elements or aspects. This aspect-focused sentiment analysis approach using LSTM-based sentiment analysis contributes to a comprehensive understanding of movie reviews. The fusion of CF with opinion mining techniques in movie recommendation systems represents a significant advancement toward enhancing recommendation accuracy and personalization.