MODEL OF MULTIAGENT SYSTEM FOR SEMANTIC TEXT ANALYSIS

Abstract

The approach to the deep text analysis and mining information on the basis of knowledge about the model of lexical language is proposed. A model for describing the process of extracting information using a system of text processing is proposed. This model enables parallel processing of text documents. With this system, we can improve the process that analysis a text document as a whole, and it does execute semantic analysis as well. The most significant advantage of using a multiagent system is the ability to simultaneously process  a text document, and this system can also help to remove repetitions from the text. The downside is that during the process, the algorithm generates multiple agent conversations, as well as breaking existing connections and establishing new ones. This behavior of the model requires a considerable amount of computing resources. The model receives a text document. The result of the model is the object coverage of the text. The set of information objects received is subsequently refined and a resulting set of objects is formed that describes the content of the document in terms of the ontology of the subject area. All the knowledge used in this approach is, to one degree or another, based on a domain model that captures the concepts and relationships of interest to the user of the system in the form of an ontology. Thus, the ontology determines what kind of information should be extracted from the available data sources. The results of each stage of processing are projected onto text, which allows to interpret the obtained results clearly and to distinguish fragments that are contextually related to each element of the received information.