NLP based automatic classification system for analytico synthetic scheme
Pijushkanti Panigrahi, A.R.D. Prasad, Abhishek Basu · SRELS Journal of Information Management · 2003
Researchers had made numerous attempts to design powerful automatic classification system, but those could not bring any well-accepted results. The main problem was successful automatic analysis of title of documents and finding out subject propositions. Because being a purely mental process, classification demands human intelligence for analysing the title to find out its basic subject and other facets, if any, along with its category and also synthesising those facets according to principles, postulates and canons to construct classification number. In other words, the document title, which is in natural language, is analysed carefully to pick up relevant words (i.e. subject propositions) and those are synthesised using classifiers expertise to build the classification number. Emergence of Artificial Intelligence (AI) could bring the solution to this problem. Use of Natural Language Processing (NLP) techniques would help in automatic analysis of titles and an Expert System be developed to work exactly in the same way as a classifier does to build classification numbers being guided by canons, principles and postulates. This paper is based on the research work in this line of thinking. Semantic and syntactic components of an AI based system for automatic classification is described. The methodology of building lexicon, parser based on Definite Clause Grammar, knowledge-base based on frame based and rule based knowledge representation model is given with a brief discussion on those knowledge representation model. Needs of developing inference engine is also discussed with example. Shows the way an expressive title is analysed to pick up noun phrases.