MEME - a society of virtual characters in an adaptive information sharing system

Messaging systems are essential for circulating information within an organization, yet currently the e-mail-based exchange of information does not take advantage of technologies that could change many aspects of the way we communicate. Here we focus on the MEME system that supports information-sharing communities. Within the messaging system each community or individual users are represented as virtual characters or avatars that react to the semantics and other features of the content being shared. Our goal is to develop models guiding the behavior of virtual characters that express such features of information as semantic relevance, timeliness or even intentions of system users. Machine learning techniques are applied to generate models of information and control the behavior of virtual characters that populate the system.