An Artificial Intelligence Technique for Information and Fact Retrieval: An Application in Medical Knowledge Processing (Information Systems)

Combining the strengths of human cognition and computer power, this book presents a practical application of artificial intelligence to the problem of retrieving information and facts with high-quality measures of precision and recall. It describes the System for Heuristic Retrieval of Information and Facts (SHRIF), a powerful and versatile system that is designed to exploit the lessons provided by humans who are able to attain partial, conflicting, and fuzzy goals effectively, despite limitations in speed and precision. In addition to information retrieval, SHRIF can serve as a model for any system with a variety of information processing objectives, from diagnosis and troubleshooting to classification and causal reasoning.

Findler employs the methodologies of knowledge representation, clustering techniques, natural language understanding, and information and fact retrieval to make use of the lessons provided by human effectiveness and flexibility in a variety of fuzzy domains. These lessons enable systems to assimilate new information, respond quickly to varying needs for information and facts, find sensible compromises in conflicting situations, distinguish between what is more and less important and between correct and faulty pieces of information, make inductive judgments on the basis of disjoint observations, discover and use partial similarities and analogies, and detect and correct errors.

Nicholas V. Findler is Research Professor of Computer Science and Director of the Artificial Intelligence Laboratory at Arizona State University.