A Logical Information Retrieval Model Based on a Combination of Propositional Logic and Probability Theory
Justin Picard, Jacques Savoy · Studies in fuzziness and soft computing · 2000
In addition to working with large amount of documents, information retrieval has to deal with the uncertainties that confront all natural languages, including homonymy, synonymy and polysemy. These represent major hurdles that every automatic natural language processing situation or system must deal with. In order to encourage the discovery of better solutions to these hurdles and to facilitate improved understanding of the matching process between query and documents, our approach is to view retrieval mechanisms as an inference process that involves uncertainty. In this context, a fundamental question involves the choice of an adequate framework which will allow us to: (1) represent the various types of uncertain knowledge; (2) combine various sources of evidence about query or document content; and (3) come up with efficient and sound techniques capable of making the needed inferences. To meet these criteria, we suggest using probabilistic argumentation systems (PAS) which combine propositional logic with probability theory such that we can deal with uncertain knowledge in both a symbolic and a numerical way. In this chapter, a model of information retrieval based on PAS will be presented. This model provides an original interpretation of van Rijsbergen’s logical uncertainty principle, a foundation for most logical IR models. Also presented will be an example of our logical model that takes hypertext links or other interdocument relationships into account, in order to enhance retrieval effectiveness. These 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.