Historically, recommender systems have primarily been concerned with recommending items to users, for example, books and movies. More recently, explicit social recommender systems have emerged, which can be categorized as social-matching (i.e., recommending individuals to each other either online or face-to-face) or social-interaction-space matching (recommending various social venues such as club meetings, political events, or online chat rooms). Researchers have begun to lay clear intellectual foundations for the exploration of the former, but not the latter. In fact only a handful of these systems have been deployed and none has used real-time models or hybrid recommendations of physical or virtual spaces.
The proposed project will examine the utility of various social-interaction-space recommendation system designs through rapid prototyping and evaluation in both the field and laboratory. This core components model will be achieved through an exploration of the character of chat-channel recommendations for large complex online (AustNet IRC network) and physical (NJIT campus) spaces. The ultimate goal is to build and sustain community that will result in the development of design guidelines and a theoretical framework that will inform system developers and HCI/CSCW researchers.
Broader Impacts Using computer technology to improve people's navigation of their social environment so that they can easily coordinate with others in activities of interest is a simple way to improve social connectivity and social capital. This is an important concern, as individuals embedded in richly connected social environments are better able to handle personal setbacks such as financial failures and illness, provide social support for others, and advance their career. Local communities and neighborhoods that have high levels of social interaction are more likely to engage in collective action, and support economic development. Systems that increase such interactions are therefore of enormous social value. This SGER will enable the exploration of a largely untested way to improve social navigation, namely deploying synchronous (near real time) social-interaction-space (chat rooms, meeting rooms, clubrooms, lecture halls, etc.) recommender systems. The project will also benefit the research community through the release of usage data from the field trials (in a suitably sanitized form) and open source software applications. Finally, the novel profiling matching algorithms and privacy mechanisms developed for this class of systems, and understanding of determinants of use of such systems, will be applicable in the government and business sectors, as well as in academia.
