A Relational Data Model for Manipulating Probabilistic Knowledge

Abstract

Reasoning with uncertain knowledge is an activity that arises in many areas of computer science, including knowledge-base systems and expert systems. It is well known that existing expert systems handle reasoning with uncertainty mostly with ad hoc techniques with little theoretical justification [12, 18]. Many systems of reasoning with uncertainty have been developed in recent times, including fuzzy set based approaches (e.g., see Zadeh [20], Raju and Majumdar [15]), quantitative approaches to reason with knowledge (e.g., van Emden [19] and Steger et al [18]) and probabilistic approaches (e.g., [7], Halpern [9]), and the classical works (for example see Carnap [3]). A highlight in the area of reasoning with probabilistic knowledge is the recent proposal by Ng and Subrahmanian [13] of a comprehensive theory of probabilistic logic programming. While these works provide tools for reasoning with uncertain knowledge, a closely related area of reasoning with and manipulating an uncertain database has received scant attention. This area is of fundamental importance as there are many database applications where available information is uncertain (e.g., stock market predictions, medical applications, to name a few). There is thus a definite need for a formal model of databases for representation, manipulation, and reasoning involving uncertain information.KeywordsProbabilistic RelationRelational AlgebraClassical RelationUncertain InformationRelevant RuleThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.