Abstract
The introduction of automated information retrieval (IR) systems was met with great enthusiasm and predictions that manual literature searching soon would be replaced. Three decades later, IR systems have not progressed to the stage where any but the dedicated few can operate them without a highly skilled human intermediary acting as interface between user and system. In the interim, we have learned that the retrieval process is extremely complex both in terms of understanding people and their communication and in terms of understanding scientific information and technical vocabulary. Experiments with new techniques suggest to many the possibility of eliminating the human intermediary, either in large part or altogether; others would argue that the retrieval problems are too complex to be resolved for more than highly restricted domains. The possibility of eliminating the human intermediary is of current research interest to the several disciplines that are represented on this panel.
The discussion will revolve around one central question: Is our inability to get end-users to search primarily a problem of understanding people and their communication, or is it primarily a problem of understanding scientific information and technical vocabulary? Among the constituent issues the panel will address are these:
Which are the most fundamental problems in constructing retrieval interfaces for the information seeker?
How successful have been the attempts to build end-user oriented interfaces?
Is a single, elegant solution to the problems of language handling, query negotiation, user modeling, and information retrieval possible?
What techniques hold the most promise for automating the functions of the intermediary?
What approaches are likely to be fruitless?
Under what conditions is it possible and appropriate to automate some or all of the intermediary's skills and function?