Integrating contextual sentiment analysis in collaborative recommender systems

Area of Science:

• Artificial Intelligence
• Information Retrieval
• Natural Language Processing

Background:

• Recommender systems are crucial for e-commerce but face challenges like cold-start and data sparsity.
• Sentiment analysis extracts user preferences from text but struggles with domain sensitivity in multi-domain applications.

Purpose of the Study:

• To propose a sentiment-based recommender system model that incorporates contextual information.
• To address the domain sensitivity issue in sentiment-based recommendation systems.

Main Methods:

• Developed a sentiment-based model integrating contextual information for recommender systems.
• Introduced a novel approach to handle domain sensitivity using contextual sentiment analysis.
• Evaluated the model through experiments comparing standard rating, standard sentiment, and contextual information models.

Main Results:

• The contextual information sentiment-based model outperformed the standard rating and standard sentiment models.
• The proposed model showed better performance compared to traditional collaborative filtering approaches.

Conclusions:

• Contextual information significantly enhances sentiment-based recommendations.
• The developed model offers a promising solution for domain-sensitive recommender systems.