User Interactions in Social Networks and their Implications

Abstract

Social networks are popular platforms for interaction, com- munication and collaboration between friends. Researchers have recently proposed an emerging class of applications that leverage relationships from social networks to improve security and performance in applications such as email, web browsing and overlay routing. While these applications of- ten cite social network connectivity statistics to support their designs, researchers in psychology and sociology have re- peatedly cast doubt on the practice of inferring meaningful relationships from social network connections alone. This leads to the question: Are social links valid indicators of real user interaction? If not, then how can we quantify these fac- tors to form a more accurate model for evaluating socially- enhanced applications? In this paper, we address this ques- tion through a detailed study of user interactions in the Facebook social network. We propose the use of interaction graphs to impart meaning to online social links by quanti- fying user interactions. We analyze interaction graphs de- rived from Facebook user traces and show that they exhibit significantly lower levels of the “small-world” properties shown in their social graph counterparts. This means that these graphs have fewer “supernodes” with extremely high degree, and overall network diameter increases significantly as a result. To quantify the impact of our observations, we use both types of graphs to validate two well-known social- based applications (RE [Garriss 2006] and SybilGuard [Yu 2006]). The results reveal new insights into both systems, and confirm our hypothesis that studies of social applica- tions should use real indicators of user interactions in lieu of social graphs.

1. Introduction

Social networks are popular infrastructures for communica- tion, interaction, and information sharing on the Internet. Popular social networks such as MySpace and Facebook provide communication, storage and social applications for hundreds of millions of users. Users join, establish social links to friends, and leverage their social links to share con- tent, organize events, and search for specific users or shared resources. These social networks provide platforms for or- ganizing events, user to user communication, and are among the Internet’s most popular destinations.

Recent work has seen the emergence of a class of socially- enhanced applications that leverage relationships from so- cial networks to improve security and performance of net- work applications, including spam email mitigation [Garriss 2006], Internet search [Mislove 2006], and defense against Sybil attacks [Yu 2006]. In each case, meaningful, interac- tive relationships with friends are critical to improving trust and reliability in the system.

Unfortunately, these applications assume that all online social links denote a uniform level of real-world interper- sonal association, an assumption disproven by social sci- ence. Specifically, social psychologists have long observed the prevalence of low-interaction social relationships such as Milgram’s “Familiar Stranger” [Milgram 1977]. Recent research on social computing shows that users of social net- works often use public display of connections to represent status and identity [Donath 2004], further supporting the hy- pothesis that social links often connect acquaintances with no level of mutual trust or shared interests.

This leads to the question: Are social links valid indi- cators of real user interaction? If not, then what can we use to form a more accurate model for evaluating socially- enhanced applications? In this paper, we address this ques- tion through a detailed study of user interaction events in Facebook, the most popular social network in the US with over 110 million active users. We download more than 10 million user profiles from Facebook, and examine records of user interactions to analyze interaction patterns across large user groups. Our results show that user interactions do in fact deviate significantly from social link patterns, in terms of factors such as time in the network, method of interaction, and types of users involved.

DOWNLOAD...