Linguistic Simulation of Semantic Invariants for Multilingual Knowledge Management Systems

Introduction The problem addressed in this paper is establishment of semantic invariants to serve as a kind of metalanguage of ”senses” valid for the natural language systems under consideration. We see the key objective of natural language processing in developing multilingual facilities of computer access to knowledge contained in texts. Our experience in design and implementation of natural language processors for knowledge-based systems allows us to assert that the focus of linguistic research is simulation of semantic-syntactic invariants [1-3]. For that we employ both theoretic study (contrastive analysis) of a subset of European languages (at present Russian, English and Italian) and computer experiment. The main hardships in NL ”understanding” by a computer program lie not so much in lexical semantics but in the semantics of structures. The given paper states out basic conclusions of our latest work in linguistic simulation for the development of the natural language knowledge-based system IKS. At present our research efforts are continued in the going on innovative project dealing with multilingual document handling. The focus of the given paper is consideration of semantic invariants which result from experimental linguistic simulation and theoretic study. The proposed solutions are deployable in knowledge management systems of various kinds. The insistent urge to develop multilingual access to knowledge contained in natural language (NL) texts (presented in electronic form) dictates the use of contrastive study for two and more languages. Contrastive study of the language systems enable us to single out the most general features which are characteristic for all the considered languages and, therefore could be a source for some basic semantic model invariant for all these languages. The current state of the