23 23 23 23 23 Journal of Database Administration Spring 1991 When processing a query in a conventional database, a set of facts (extensional answers) are usually returned for an answer set. This can be different in a deductive or expert database system. Users may, as before, be interested in a set of facts to answer their query, but in certain queries users are also able to get the answer of a query as a set of formulas (intensional answers). These “intensional answers” can greatly reduce the costs of processing a query, can be represented in a more compact way than a large set of facts, and are moreover independent of the current state of the database. In this paper, we introduce an intensional query processing technique composed of three steps and its im- plementation in Prolog. The three steps are pre-resolution, resolution, and post-resolution. In pre-resolution step, deductive rules are transformed into certain forms. In resolution step, SLD- resolution is applied until the last resolvents consist of either extensional literals or comparison literals. In post-resolu- tion step, intensional answers are generated and checked against integrity constraints to remove meaningless an- swers. Our work is an extension to Cholvy and Demolombe (1986), and Pascual and Cholvy (1988). Introduction and Motivation Introduction and Motivation Introduction and Motivation Introduction and Motivation Introduction and Motivation Relational database systems generate only a set of factual data (extensional answers) as an answer set for a given query. Deductive database systems, however, can generate a set of first order logic formulas as an answer set (intensional answers) for a given query. A query is called intensional if answers to the query can be represented by a set of formulas. When querying a deductive or expert database database for intensional answers, we are not interested in the set of objectsa that satisfy a given query in a particular database state. Instead, we want to know the conditions that objects must satisfy, in any state, to belong to an answer. But why are we dealing with intensional query proc- essing at all? As we get intensional answers as a set of formulas, they are independent of the particular circum- stances in the database. Displaying the output of a query in terms of a formula gives us the answer in a more compact form than a set of facts could ever do. Intensional query processing also has an advantage in computation compared with extensional answers. Intensional answers can be com- puted without accessing the database, which saves a lot of time. Also, intensional answers tell us exactly what condi- tions must be fulfilled to get a certain extensional answer. That means intensional answers can help us to interpret extensional answers. This paper looks closely at some work dealing with the topic of intensional answers and suggests a technique com- posed of three steps. Terminology Terminology Terminology Terminology Terminology We assume the reader is familiar with the terminology of first-order logic which are normally defined in standard logic references (Chang & Lee, 1973), such as formulas, Design and Implementation of a Three-Step Intensional Query Processing Scheme Il-Yeol Song Drexel University Hyoung-Joo Kim Seoul National University Manuscript originally submitted December 3, 1990; Revised March 8, 1991; Accepted April 30, 1991 for publication.