Vladimir A. Kulyukin, Kristian J. Hammond, Robin D. Burke
The World Wide Web continues to challenge organizations to make online access to their expertise convenient for their clients. One means of expertise access that many clients find convenient in everyday life is asking natural language questions of the organization. To support it online, we developed an approach to building organization-embedded question-answering intermediaries, called Information Exchange systems. These systems use their knowledge of the organization’s structure to answer the clients’ questions and to acquire new expertise from the organization’s experts. Our approach uses techniques of hierarchical and predictive indexing, combined term weighting, abstraction-based retrieval, and negative evidence acquisition. We illustrate our approach with the Chicago Information Exchange system, an Information Exchange application embedded in one university’s computer science department.