Automatic audience focusing by event interestingness

Abstract

Large social Networks have marketing potential to spread information about interesting events to suitable audiences. However, huge network sizes and varieties of information available are obstacles to reach the desired goal. This paper investigates the hypothesis of computable Interestingness as a criterion to focus on suitable audiences for any given event. Interestingness is calculated by combining two functions: Relevance and Surprise. A generic software tool has been developed as an experimental testbed to interact with any social network. Its inputs are the event characterization and audience candidates for the given event. Two results validate this work's hypothesis: first, audience candidates who actually visited the event site, have on the average a bigger computed Interestingness than the rest of the population; second and most important, computed Interestingness better differentiates event site visitors, actually interested in the Event, from non-visitors, while Relevance alone, does not distinguish so-well between visitors and non-visitors. 1.