دانلود رایگان مقاله انگلیسی رفتار تکرار جمعی برای شبکه های اجتماعی آنلاین - الزویر 2018

Abstract

Understanding the patterns of collective behavior in online social network (OSNs) is critical to expanding the knowledge of human behavior and tie relationship. In this paper, we investigate a specific pattern called social signature in Facebook and Wiki users’ online communication behaviors, capturing the distribution of frequency of interactions between different alters over time in the ego network. The empirical results show that there are robust social signatures of interactions no matter how friends change over time, which indicates that a stable commutation pattern exists in online communication. By comparing a random null model, we find the that commutation patter is heterogeneous between ego and alters. Furthermore, in order to regenerate the pattern of the social signature, we present a preferential interaction model, which assumes that new users intend to look for the old users with strong ties while old users have tendency to interact with new friends. The experimental results show that the presented model can reproduce the heterogeneity of social signature by adjusting 2 parameters, the number of communicating targets m and the max number of interactions n, for Facebook users, m = n = 5, for Wiki users, m = 2 and n = 8. This work helps in deeply understanding the regularity of social signature.

4 Conclusion and discussions

In this paper, we empirically investigate collective behavior of online users’ social signature by means of analyzing ego network and regenerate the heterogeneity of social signature. FBWall and Wikitalks are divided into 4 equal length time intervals. In each interval, we construct ego network in which each ego is tied at least 15 alters. Collective egos are those who appear in all 4 intervals with social signature. The statistical properties of frequency of interactions of ego network are measured by social signature P, persistence JSD, turnover J and slope λ. Empirical results show that: 1) persistence of social signatures exists over time no matter what turnover of alters; 2) social signatures are heterogeneity by comparing the null model. Furthermore, we present a preferential interaction model in term of weighted network, which include two mechanisms: 1) New node follow topological growth according to the probability of nodes strength; 2) Old nodes follow preferential interaction according to the order of nodes, while in each time step, an old node reaches m existing nodes and interact randomly from 1 to n times. The simulation results show the high similarity of social signature regenerated by preferential interaction model with empirical results. One can find that 1) The strength of nodes causes the new coming user tend to choose the old user who has strong social tie; 2) The recency of nodes causes the old user tend to communicate with the new users. Furthermore, the heterogeneity of social signature can be adjusted by 2 parameters, the number of communicating targets m and the max number of interactions n, for Facebook users, m = n = 5, for Wiki user, m = 2 and n = 8.