Using multiwinner voting to search for movies.

Area of Science:

• Information Retrieval
• Computational Social Choice
• Recommender Systems

Background:

• Traditional recommendation systems often lack control over result diversity.
• Multiwinner voting offers a flexible framework for preference aggregation.

Purpose of the Study:

• To present a prototype system utilizing multiwinner voting for resource recommendations.
• To demonstrate how different voting rules can control the diversity of suggested resources.
• To evaluate the system's performance on both synthetic and real-world datasets.

Main Methods:

• Implementation of a prototype recommendation system based on multiwinner voting.
• Application of various voting rules to a query set of user preferences.
• Testing and validation using synthetic data and the MovieLens dataset.
• Visual comparison of search results across different diversity levels.

Main Results:

• The system successfully generates recommendations closely related to the query set or a broader spectrum of options.
• The choice of voting rule directly influences the diversity of the recommendations.
• The system's effectiveness is validated on both simulated and real-world movie rating data.

Conclusions:

• Multiwinner voting provides an effective mechanism for controlling recommendation diversity.
• The developed system offers a tunable approach to resource suggestion, balancing relevance and variety.
• This approach has implications for personalized content delivery and information discovery.