APPLICATION OF SEMANTIC NETWORK FOR SEARCH OF NEW TERMS AND PATTERNS IN MULTIMODAL AIS DATA BASED ON ARTIFICIAL INTELLIGENCE

Yurchak Vladimir Alexandrovich

With the introduction of artificial intelligence into the analysis and processing of multimodal data, it has become possible to extract information from unstructured texts, to perform automatic clustering, Categorization and classification of information with subsequent use in various scientific and production sectors. Automated analysis allowed researchers not only to collect and study the amount of material that could not be analyzed manually, but also to identify patterns and relationships between different terms. Automated analysis based on international computer systems, The dictionary and database set itself the task of searching for new concepts and terms in already analyzed multidimensional data sets. This article addresses the problem of automated analysis of multimodal data, namely: identification of common terms and concepts at the junction of one or more subject areas (thematic clusters). In particular, the study describes approaches to building a semantic network at the intersection of several subject areas (thematic clusters) based on the terms of the international knowledge base of medical publications (PubMed) and the international dictionary of medical terms (UMLS). The approaches to building a semantic network at the intersection of several subject areas (thematic clusters) considered in this article include mechanisms for forming thematic clusters, The identification of semantic links between terms and the visualization of new terms. The result described in this article is the creation of a semantic network (graph) «Diseases and their mutations» on the basis of PubMed libraries (PMC) and UMLS, which will help to define new terms for disease mutations and understand the reasons why they occur.

Keywords:semantic network(graph), multimodal data, automated analysis, UMLS, PubMed, BERTopic, thematic clusters