The worldwide expansion of network access is driving an increase in interactions among people, among businesses, and between people and businesses. Such interactions are the staple of major Internet uses such as electronic commerce and virtual communities. Successful interaction relies heavily upon trust. Whereas security seeks to prevent illegal actions, trust goes beyond security in seeking assurance of accountability (of intent and capability) even for legal actions. However, figuring out whom to trust and to what extent is extremely difficult in an open networked environment such as the Internet. This project will study distributed, scalable computational approaches for trust management taking into account the different forms in which trust is exhibited in networked environments. Such approaches rely less on centralized authorities and more on community policing through reputation mechanisms. This project will evaluate the approaches based on criteria such as how easily a given approach can be bootstrapped, how efficient it is in helping members find competent and good peers, and how immune it is to invasion by untrustworthy members. This project will also study aggregate phenomena such as the emergence of subcommunities, linkages among subcommunities, and the sensitivity of a community to change in membership.
