Where do I start?

Hao Wu, Michael Mampaey, Nikolaj Tatti, Jilles Vreeken, M. Shahriar Hossain, Naren Ramakrishnan
2012 Proceedings of the ACM SIGKDD Workshop on Intelligence and Security Informatics - ISI-KDD '12  
The "where do I start?" problem is a veritable one in intelligence analysis. We identify several classes of algorithmic strategies that can supply starting points to analysts in their exploration of a document collection. We present nine methods with origins in association analysis, graph metrics, and probabilistic modeling, and systematically evaluate them over multiple document collections. One of these methods, a novel approach to modeling "surprise", is our specific contribution and,
more » ... , supports the iterative refinement of suggestions based on user feedback. We demonstrate how these methods guide the analysts to start their investigation on intelligence document collections. Our results reveal selective superiorities of the algorithmic strategies and lead to several design recommendations for creating document exploration capabilities.
doi:10.1145/2331791.2331794 fatcat:tnnlnbdwdba5di2hmqsnx3kjru