Word Agent Based Natural Language Processing

Natural language processing (NLP) is often based on declaratively represented grammars with emphasis on the competence of an ideal speaker/hearer. In contrast to these broadly used methods, we present a procedural and performance-oriented approach to the analysis of natural language expressions. The method is based on the idea that each word-class can be connected with a functional construct, the so-called word agent, processing its own part of speech. The syntactic-semantic analysis performed by the word-class agents is surveyed by a WordClass Agent Machine (WCAM) used in the bibliographic information retrieval system LINAS at the University of Hagen. The language processing is organized into four main levels: on the rst and second level, elementary and complex semantic constituents are created which have their correspondence in mental kernels built during human language understanding. On the third level, these kernels are used to construct the semantic representation of the propositional kernel of a sentence. During this process, the subcategorization features of verbs and prepositions play the main part. The task of the last level consists in the syntactic analysis and semantic interpretation of modal operators in the broadest sense and of the coordinating or subordinating conjunctions which connect the propositional kernels.