AAAI-96 Workshop on Internet-Based Information Systems

Article excerpt

The Workshop on Internet-Based Information Systems, held as part of the Thirteenth National Conference on Artificial Intelligence, provided a venue for researchers to discuss the role of Al and related fields in network-based information systems. Not surprisingly, most of the discussion focused on the World Wide Web (WWW), and a good part of the workshop was devoted to topics related to WWW search and retrieval. A number of presentations addressed ways to improve search accuracy and efficiency. Methods under discussion included statistical modeling, machine-learning techniques, and the addition of ontologically categorized metadata to WWW pages. An invited talk by Michael Mauldin of Lycos Inc., who described the development, commercial deployment, and current state of the LYcos search engine, provided an industrial perspective on this topic. It seems certain that research in this area will continue to expand. At the same time, the successful transfer of research results to commercial search engines will require persuasive costbenefit analyses for new proposed methods.

A second major theme was the use of natural language-processing techniques for analyzing WWW pages. For example, two presentations described means of taking advantage of the question-answer format of frequently asked question (FAQ) files to provide more precise information-retrieval abilities (albeit limited to the narrowly circumscribed domains covered by FAQ files). To address the bottleneck of manually creating hypertext from linear documents, another presentation described a technique for the automatic creation of hypertext links between related passages. …