Integrating multi-criteria decision-making with hybrid deep learning for sentiment analysis in recommender systems

Area of Science:

• Artificial Intelligence
• Information Retrieval
• Decision Science

Background:

• Expert assessments in decision-making models are often limited by predefined numerical or language terms.
• Information overload online necessitates intelligent decision-support technologies like recommender systems.
• Traditional recommender systems rely on single-grading algorithms, limiting nuanced user preference capture.

Purpose of the Study:

• To develop a method that incorporates expert judgments expressed in natural language into decision-making models.
• To enhance recommender systems by analyzing expert reviews and star ratings for multi-criteria decision making.
• To address information overload by providing more informed user choices through sentiment analysis.

Main Methods:

• Proposed the Sentiment Analysis in Recommender Systems with Multi-person, Multi-criteria Decision Making (SAR-MCMD) method.
• Utilized sentiment analysis to process natural language expert reviews and star ratings.
• Integrated deep learning with multi-criteria decision-making for enhanced recommendation capabilities.
• Validated the technique using datasets from TripAdvisor, TMDB 5000 movies, and Amazon.

Main Results:

• Achieved a sentiment analysis accuracy of 98% for expert reviews.
• Demonstrated superior performance in accuracy, precision, recall, and F1 score compared to existing methods.
• The SAR-MCMD system provides highly accurate suggestions to users.

Conclusions:

• Sentiment analysis of natural language expert reviews can significantly improve decision-making models.
• Combining deep learning with multi-criteria decision-making offers a powerful approach for advanced recommender systems.
• The SAR-MCMD method effectively handles complex expert judgments for better recommendations.