Accurate POI recommendation for random groups with improved graph neural networks and a multi-negotiation model

Area of Science:

• Computer Science
• Artificial Intelligence
• Human-Computer Interaction

Background:

• Group activities are increasing, driving demand for group Point of Interest (POI) recommendations.
• Current methods for fixed groups are advanced, but personality-aware recommendations for random groups remain underexplored.
• Existing systems provide POI lists, leading to suboptimal group choices and poor user experience.

Purpose of the Study:

• To address the gap in personality-aware POI recommendations for random groups.
• To improve the user experience by directly recommending an optimal POI, rather than a list.
• To propose a novel model, Accurate POI Recommendation for Random Groups with improved Graph Neural Networks and a Multi-negotiation Model (APRRGM).

Main Methods:

• APRRGM generates group features using member personalities and POI interaction data.
• It employs improved Graph Neural Networks (GNNs) to learn POI features, incorporating member personalities.
• A multi-negotiation model determines the optimal POI from a recommended sequence based on group features and POI characteristics.

Main Results:

• Extensive experiments were conducted on Yelp, Gowalla, and Foursquare datasets.
• APRRGM demonstrated superior performance compared to existing baseline models.
• The model effectively integrates personality traits into the recommendation process for random groups.

Conclusions:

• APRRGM offers an accurate and improved approach to POI recommendation for random groups.
• The integration of personality-aware features and a multi-negotiation strategy enhances recommendation quality and user satisfaction.
• This work advances the field of personalized group recommendation systems.