Causality for purposes of automatic retrieval

This paper aims to illustrate one way in which linguistics may contribute to present-day knowledge management by describing the framework and working of the sofware which the URUK-project has developed for purposes of document management. The various parts of the tool which will be described are centered around two different functions for natural language processing. One is the semi-automatic method for thesaurus construction. This formed the point of departure for the development of the tool's main function, viz. a parsing system for causal relations between relevant thesaurus terms, including identification of actual causes and results. In order to situate the development and working of the tool for causal retrieval optimally, section 3 of the article sets out with a discussion of the components which constitute a general definition of causality and the functional levels of language use where such a relation may be established (3.1). Taking the general description further, we propose a subclassification of the concept of causality with two variable parameters. In this way we have been able to narrow down a semantic working definition of causality, which was manageable for electronic retrieval in a first stage (3.2). After establishing our semantic field of action in this manner, important practical decisions for the tool's linguistic input are explained (3.3). The tool's output potential is then briefly demonstrated (section 4). Finally, two linguistic aspects for expanding and improving the current tool are discussed with relation to characteristics of natural language (section 5).