User Participation in Collaborative Filtering-Based Recommendation Systems: A Game Theoretic Approach.

Area of Science:

• Computer Science
• Artificial Intelligence
• Game Theory

Background:

• Collaborative filtering is essential for recommendation systems, but relies on user ratings.
• Users often minimize ratings due to cost, creating a tradeoff between rating effort and recommendation quality.

Purpose of the Study:

• To model user interactions in recommendation systems as a game in satisfaction form.
• To introduce and analyze the concept of satisfaction equilibrium (SE) in this context.
• To develop and evaluate a behavior rule for users to reach SE.

Main Methods:

• Modeling user interactions as a game in satisfaction form.
• Defining and analyzing satisfaction equilibrium (SE).
• Designing and theoretically analyzing a user behavior rule for iterative rating.
• Conducting experiments on Jester and MovieLens datasets.

Main Results:

• A satisfaction equilibrium (SE) can be achieved under specific conditions through iterative rating.
• Experimental results validate the theoretical analysis.
• Users can obtain satisfying recommendations with minimal ratings if expectations are moderate and satisfied users contribute more.

Conclusions:

• The satisfaction equilibrium (SE) framework offers insights into balancing user participation and recommendation quality.
• The proposed behavior rule facilitates achieving SE, encouraging user engagement.
• Findings aid in designing effective mechanisms to boost user participation in recommendation systems.