The predictability of social trends depends on our knowledge of the mechanisms of social influence and information diffusion. On the one hand, social influence is related to the expertise and structural position of the user in the network. On the other hand, information diffusion and social contagion are related to the topic of disseminated content and social influence of nodes that it passes through. Our studies contribute to both of these aspects. We propose and compare various ways of measuring social influence and analyze information cascades. Finally, we study how various mechanisms combined together shape the emergence of social conventions.

• Measuring social influence
• Diffusion of information
  
• Emergence of social conventions
  

 

Online Social Networks (OSNs) are increasingly being used as a 'media' to share and find recent information on the topics of one's interest. The user-generated content in OSNs constitutes a tremendously rich source of information, which users are increasingly relying on to discover interesting news. To facilitate it, information retrieval systems e.g., search and recommender systems, are deployed in all popular OSNs today. However, such systems in OSNs are still in their infancy compared to information retrieval systems deployed on the Web. Our research focuses on characterizing users, e.g., inferring their topics of expertise and interest, analysing how users form groups and what kinds of groups they form, and developing improved search and recommendation systems in OSNs.

• Discovering topical experts in the Twitter social network
• Exploring topical and social groups in social networks
• Random vs. expert sampling of the Twitter streams
• Geographic dissection of the Twitter Network

 

While Online Social Networks (OSNs) like Facebook, Google+ and Twitter have succeeded in creating a large user base, the service providers as well as users of these services are still faced with a variety of security and privacy issues.

Security: One of the key security threats plaguing social systems is abuse of services. Service abuse takes many forms -- e.g., users are flooded with spam messages for the purpose of promoting businesses or for spreading malware, and crowd-sourced content rating services are tampered by fake or Sybil accounts to manipulate ratings. Such attacks severely degrade the quality of service provided by such social networking websites. Our work tries to understand existing attacks and defenses, and propose novel techniques to build robust and practical systems to limit service abuse.

Privacy: Users today are uploading large amounts of content that are personal in nature. Such sensitive pieces of content are shared with the OSN provider, third party applications, and other users of the services. This has led to serious privacy leakage issues and also increased the privacy management overhead for the user. We perform large-scale measurement studies to understand privacy risks with sharing personal content online and propose new systems to provide strong privacy guarantees.

Limiting service abuse

• Limiting service abuse by Sybil identities
  
• Understanding spammer infiltration in social networks and defending against collusion attacks

Understanding privacy issues

• Understanding privacy leakage via the social network
• Understanding the user privacy management overhead

Building systems with better privacy

• Limiting privacy leakage via third party apps
• Towards decentralized OSNs for better control over personal data
• Simplifying Access Control specification in OSNs
• Towards better user privacy via exposure control