Logics and Rules

Abstract

Logic-based systems have long been used in the retrieval of documents and bibliographic records. The performance of early manual systems, and more recently automated retrieval systems, can be determined using the analytic methods described above. These inherently quantitative methods may be applied, in many cases, to logical queries because of the set-theoretic basis for logic and for probability theory. Since we can determine the size of sets associated with variables referred to in logical statements, we can compute the frequencies and probabilities needed for estimating performance. In some cases logic based retrieval will produce search results for the user inferior to results produced by a system using the probabilistic model that allows for varying probabilities of relevance. However, logical systems are relatively easy to implement and may retrieve documents faster than probabilistic systems, making logic-based systems useful in many situations.

Nothing is as practical as good theory.

—Harpers, December 1993