Modern systems embrace a multiagent setting where a system consists of multiple autonomous parties. These individual parties have their own  goals and have to achieve them by interacting with each other due to  their limited capabilities and resources. Trust provides a basis for  effective interaction, helping agents decide which parties to rely  upon for taking a particular action. My research aims to (1) build  trust models that enable agents to estimate trustworthiness of parties  based on (limited) evidence, and (2) develop approaches for agents to  make effective predictions and decisions by mining trust relations.  In this talk, I will introduce the general trust model, and talk about  three applications on social networks. CertProp and Shin are two  approaches that estimate trustworthiness of unknown parties by  propagating trust through existing trust relations. LocPat is a trust  mining framework providing flexible, personalized recommendations  based on trust relations. The significance of these approaches is  three-fold. My approaches build trust from an individual, ego-centric  perspective. They are decentralized approaches that only require a  local view of the whole system. They are effective compared to  existing approaches. 

Chung-Wei Hang is a postdoc in Computer Science department of North Carolina State University. He received his BS and MS of Computer and Information Science from National Chiao Tung University in Taiwan. He received his PhD in Computer Science from North Carolina State University. Chung-Wei's research interests include artificial intelligence, multiagent systems, trust models, recommender systems, social networks, and service-oriented computing.