Encyclopedia of Library and Information Science

Introduction Database systems are computerized systems for managing large amounts of data. Their principal function is to process requests from users to retrieve or alter the data they store. Traditionally, such requests must be expressed in a formal query language. Requests that are well defined are performed promptly; other requests are declined. Processing requests for information is also a frequent human activity, but in contrast with the rather rigid behavior of database systems, humans are capable of cooperation. A cooperative person will attempt to guess the information need of the requester, even when the request itself is flawed or unsatisfiable, and will then respond to this need rather than the request. Cooperation is considered to be an aspect of human intelligence. As the term suggests, the area of cooperative database systems is concerned with techniques that enhance database management systems with such cooperative behavior. These techniques usually imitate some aspect of human behavior. This review of cooperative database systems focuses on methods that imitate some aspects of human cooperation, and we note that there are many other methods that enhance the usability and responsiveness of database systems but fall outside the scope of this review. These include graphical user interfaces and query-by-example interfaces, the use of profiles to store user preferences, the extension of query languages to allow users to express vague ("fuzzy") requests, and in general, any method that decreases the skills that users must have. Often these methods are combined with cooperative methods to create database systems that are oriented toward naive users. Research in cooperative database systems is strongly indebted to research in artificial intelligence, in particular the work done by Joshi, Webber, and others at the University of Pennsylvania (1-5). The main interest of that group was natural language discourse between users and systems, with emphasis on mutual beliefs, misconceptions, and cooperation. Database researchers for the most part have deemphasized issues related to natural language understanding, and have concentrated on developing techniques for well-known data models, such as those based on relations or on mathematical logic. Our review of cooperative database systems is organized around a simple four-category classification of the various types of cooperative behavior. (See the section "Classification" below.) This classification is based on the stage in the user-system interaction process in which cooperation is introduced. The sections "Formulation of Objectives and Conversion of Objectives to Queries," "Detecting Deficiencies in Apparently Good Queries," and …