An improved memory-based collaborative filtering method based on the TOPSIS technique

Area of Science:

• Computer Science
• Information Retrieval
• Artificial Intelligence

Background:

• Recommender systems are crucial for filtering vast online data based on user preferences.
• Collaborative Filtering (CF) is a prevalent recommendation technique relying on user preference correlations.
• Existing CF enhancements focus on similarity measures, with less attention on prediction score methods.

Purpose of the Study:

• To propose and evaluate a novel prediction score method for memory-based collaborative filtering.
• To improve the accuracy of recommender systems by integrating the TOPSIS method into CF.
• To offer an alternative to existing prediction score methods for item evaluation and ranking.

Main Methods:

• Developed a recommendation approach using the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS).
• Applied the TOPSIS method as an alternative to traditional prediction score techniques in CF.
• Evaluated the proposed TOPSIS-based CF method on the MovieLens 100K and 1M benchmark datasets.

Main Results:

• The TOPSIS-based collaborative filtering method demonstrated improved recommendation accuracy.
• The proposed approach outperformed common baseline CF methods across various evaluation metrics.
• This indicates the effectiveness of TOPSIS in enhancing the prediction capabilities of recommender systems.

Conclusions:

• The integration of TOPSIS offers a significant improvement over standard prediction score methods in memory-based CF.
• This research highlights the potential of multi-criteria decision-making methods for advancing recommender system accuracy.
• The findings suggest that TOPSIS-based CF is a more accurate and reliable approach for personalized recommendations.