Information Retrieval Systems an Algebraic Approach II

Article type: Research Article

Authors: Raś, Zbigniew

Affiliations: Warsaw University

Abstract: This paper is the second of the three parts of the work on the information retrieval systems proposed by Salton (see [13]). The system is defined by the notions of a partially ordered set of requests (A, ⩽), the set of documents X and a monotonic retrieval function U : A → 2X. Different conditions imposed on the set A and a function U make it possible to obtain various classes of information retrieval systems. We investigate systems in which (A, ⩽) is a partially ordered set, a lattice, a pseudo-Boolean algebra and a Boolean algebra. In my paper these systems are called partially ordered information retrieval systems (po-systems), lattice information retrieval systems (l-systems), pseudo-Boolean information retrieval systems (pB-systems) and Boolean information retrieval systems (B-systems). The first part concerned po-systems and l-systems. The second part deals with pB-systems and B-systems. In the third part systems with a partial access are investigated. The present part discusses the pB-systems and B-systems. The problems connected with the properties of sets of attributes are investigated. The method of constructing a selective system for a pB-system and a B-system is given. The problem related to the minimalization of systems are investigated.

Keywords: information retrieval, selectivity, artificial attributes, single systems, pB-systems, B-systems

DOI: 10.3233/FI-1981-4403

Journal: Fundamenta Informaticae, vol. 4, no. 4, pp. 777-818, 1981