Principles of content analysis for information retrieval systems: an overview

Abstract

"Unquestionably, the content analysis which has emerged as part of Information Retrieval Systems (IRS, e.g. literature databases) over the past 20 years has much in common with the content analysis used by linguists or in the social sciences. However, its intrinsic value stems from the special context in which it is used: a) Close interdependencies link the selected content analysis with the retrieval situation. The user’s retrieval strategies, which are intended to obtain information relevant to the current problem situation, and the available aids (e.g. expansion lists or user-friendly browsing tools) affect the efficacy of some analysis techniques (e.g. noun phrase analysis from computer linguistics) to a considerable extent. b) Normally, a commercial IRS handles mass data, thus necessitating the use of a reduced content analysis even today. Full morphological, syntactic, semantic and pragmatic text analyses are unthinkable simply for efficiency reasons but also for knowledge reasons. Content analysis in IRS is therefore a component part of a special type of restricted system which obeys its own laws. Against the backdrop of these considerations, forms of content analysis in present-day commercial retrieval systems are studied and promising expansions and alternatives are proposed." (author's abstract)