Social Interaction Model Design for University Campus Running Programs

Abstract

To reveal the intrinsic relationship between college students' campus running behavior patterns and their social interactions, this study proposes and constructs a campus running social interaction model based on multidimensional exercise data. The study utilizes a simulated campus running dataset, comprising 50 students, 395 behavior records, and 14 physiological and behavioral indicators, including heart rate, running speed, distance, pace, cadence, energy expenditure, and recovery time. Behavioral pattern analysis methods were applied to extract students' exercise features across different time periods. Subsequently, the cosine similarity algorithm was employed to quantify behavioral similarity among students, and a campus running social network was constructed using a similarity threshold of 0.5. The results indicate that a target student can identify 19 potential high-similarity interaction peers, with the top 10 most similar students exhibiting similarity scores ranging from 0.84 to 0.97, demonstrating significant behavioral aggregation. Social network analysis further reveals that the overall network density exceeds 0.8 and the clustering coefficient is above 0.9, indicating high connectivity and group cohesion. Combined with visual analyses, including heatmaps and time series plots, the study validates the association between behavioral similarity and stable social interactions. This research provides a quantifiable and reproducible data-driven framework for designing personalized campus running intervention strategies and modeling social interactions in the context of smart campus sports management.