Alike People, Alike Interests? A Large-Scale Study on Interest Similarity in Social Networks
Xiao Han,Leye Wang,Soochang Park,Angel Cuevas,Noel Crespi
DOI: https://doi.org/10.1109/asonam.2014.6921631
2014-01-01
Abstract:This paper presents a comprehensive empirical study on the correlations between users' interest similarity and various social features across three interest domains (i.e., movie, music and TV). This study relies on a large dataset, containing 479, 048 users and 5, 263, 351 user-generated interests, captured from Facebook. We identify the social features from three types of the users' information - demographic information (e.g., age, gender, location), social relations (i.e., friendship), and users' interests. The results reveal that the interest similarity follows the homophily principle. Particularly, the results show that two users are more likely to be alike in their interests 1) if they exhibit more similarity in their demographic characteristics (e.g., similar age, same gender, or close to each other geographically), or 2) if they are more intimate in their friendship, or 3) if they present a higher average interest individuality (i.e., a measurement for estimating the personalized characteristics of a user's interests). The empirical observations could be exploited to infer how two users are alike in their interests according to the social features, which could be further harnessed by various practical applications and services, such as recommendation system and advertisement service.