Representation of semantically cohesive sentence fragments in scientific and technical texts

Abstract

The paper presents a formal sentence model of scientific and technical texts to be used in the process of semantic interpretation. The model is based on the word dependencies, formed by sequences of word combinations that are called dependency chains. Characteristic word sets are defined within the dependency chains and the discriminated transitive closure of direct dependence as well as its notation in the form of the incidence matrix are assigned. The procedures and algorithms of semantically cohesive sentence fragments detection are constructed on the basis of the introduced concepts.