ITR: The Construction and Analysis of Information Networks

Cornell University, Ithaca NY

Investigators

Abstract

Sources of on-line information are becoming increasingly decentralized, heterogeneous, and complex even as they become correspondingly richer and more valuable. Determining the structure of these information sources is becoming key to extracting and managing the knowledge they contain. Some of these sources exhibit an explicit network structure --- the hyperlinks of the World Wide Web form an excellent example. In other domains, ranging from electronic communication to informal human social networks, subtle hidden linkage relations play a large role in determining the information flow within and between communities. The link structures of both types of environments can yield a surprising wealth of latent information about their content, making their complexity manageable. The proposed research seeks effective mechanisms for eliciting a global understanding of link structures in information networks. A key component of this effort is the design of techniques and tools enabling a richer level of interaction with on-line information. The research focuses on the development and integration of new techniques in three areas: natural language understanding methods to uncover implicit relationships in on-line content, efficient algorithms to analyze complex networks of inter-connections, and mathematical models of the dynamics and social processes by which networked information evolves.