Abstract: Internet search engines have attracted widespread demand for information retrieval from unstructured documents. The number of structured and semistructured documents available on the Web is huge‹and collections of these are more amenable to data mining than search engine retrieval.
Finding patterns in databases of political contributions, environmental data, or hospital and school performance would surely interest many citizens. However, compared to search engines, there has been no similar explosion of interest in data mining. Why?
The main research question is how to support such exploration for users with little or no training in statistics or programming.
In contrast to other data mining systems, Bungee View focuses on learnability, responsiveness, robustness, and providing a satisfying user experience. This talk will describe users¹ experience with Bungee View in the lab and on three Web-based image collections.

Bio: Mark Derthick received his PhD in Computer Science from Carnegie Mellon University in 1988 for his thesis that Connectionist models of knowledge representation and reasoning would degrade more gracefully than symbolic frameworks in the face of incomplete and inconsistent information. After graduation, he went over to the symbolic side and worked on analogical reasoning with the Cyc large-scale common-sense knowledge base. He also developed user interfaces for entering knowledge into Cyc.
In 1995 he returned to CMU, working on user interfaces for exploring structured data with the Sage project. His current projects are summarizing probability distributions over tens of thousands of possible evolutionary trees for biologists, and developing an enjoyable interface for non-technical users to browse and data-mine image collections.
The Human-Computer Interaction Institute has been a rewarding environment for coordinated development of information architecture, visual and interaction design, and algorithms that support simple yet powerful interfaces for exploring these data.