An intelligent film recommender system based on emotional analysis

Area of Science:

• Artificial Intelligence
• Natural Language Processing
• Recommender Systems

Background:

• Traditional personalized film recommendations rely on user ratings, which are subjective and vary widely.
• This subjectivity makes it challenging to accurately capture user preferences and deliver relevant film recommendations.
• Mining user review sentiment offers a promising alternative for understanding true user preferences.

Purpose of the Study:

• To develop a personalized film recommendation method that leverages sentiment analysis of user reviews.
• To address the limitations of subjective user rating scales in existing recommendation systems.
• To enhance the accuracy and user engagement of film recommendations through emotional analysis.

Main Methods:

• Utilized natural language processing (NLP) to extract emotional tendencies from user film reviews.
• Implemented a weighted fusion of multi-modal emotional features, optimized using Particle Swarm Optimization (PSO).
• Calculated emotional similarity considering content publishing time and average user sentiment to generate top-N recommendations.

Main Results:

• The proposed multi-modal sentiment analysis-based recommendation system outperformed comparison methods.
• The system effectively mitigated issues arising from inconsistent user rating scales.
• Demonstrated an increase in users' interest and engagement with recommended films.

Conclusions:

• Sentiment analysis of user reviews provides a robust approach for personalized film recommendations.
• The developed method offers a superior alternative to traditional rating-based systems.
• This approach enhances the quality and user satisfaction of film recommendation services.