Experiments in using agent-based retrieval from distributed and heterogeneous databases

Intelligent agents are being widely employed by users to assist in information retrieval tasks. However, users still suffer from "information overload" and most search engines are limited, particularly in meeting user expectations. We have developed SAIRE (Scalable Agent-based Information Retrieval Engine), which addresses many current search engine limitations. SAIRE employs intelligent software agents, natural language (NL) understanding and conceptual search techniques to support Internet access to Earth and space science data maintained by NASA and NOAA at distributed technical centers across the USA. This report describes progress made to date on the SAIRE prototype development effort. It includes progress made in developing agents' persistent query handling capabilities. Our report also details the current status of SAIRE's user modeling and concept-directed search capabilities.