REPRESENTATION AND CLASSIFICATION OF UNSTRUCTURED DATA IN ADAPTIVE LEARNING MULTIMEDIA SYSTEMS BASED ON THE METHOD OF COMPARATIVE IDENTIFICATION

Abstract

The article addresses the problem of representation and classification of unstructured data in adaptive teaching multimedia systems. The relevance of the study is determined by the rapid growth of heterogeneous educational resources, including text, images, audio, video, and combined multimedia objects, whose weak formalization complicates automated processing, semantic interpretation, and personalized delivery. The purpose of the research is to develop a formal approach to organizing such resources by applying the method of comparative identification to documents selected from a multimedia database. The proposed approach is based on defining the correspondence between a document and a concept of the subject domain through a relevance predicate. On this basis, equivalence relations are introduced both for multimedia documents and for domain concepts, which makes it possible to form classes of semantically close documents and classes of functionally equivalent concepts. Such factorization reduces redundancy, preserves semantic integrity, and creates the basis for constructing a hyperstructure of educational content. The paper shows that the use of predicates as formal descriptors enables consistent classification of unstructured resources and supports semantic relationships between content elements at different levels of abstraction. In this way, the method can be incorporated into the architecture of adaptive multimedia and hypermedia systems, where the representation of knowledge, user modeling, and navigation control must operate jointly. The proposed model supports not only storage and indexing, but also adaptive presentation of learning materials according to user needs, prior knowledge, and educational goals. This is especially important for environments that rely on intelligent data analysis to generate individualized learning trajectories and to improve the accessibility of educational resources. A distinctive feature of the proposed solution is the simultaneous classification of both documents and concepts within a unified formal framework. This allows the system to organize multimedia content into semantically coherent layers, establish stable semantic relationships between learning objects, and provide multi-level navigation within the educational space. As a result, learners can receive logically ordered materials that correspond more precisely to their current tasks and cognitive characteristics. The method also creates preconditions for further automation of content processing, including the integration of machine learning tools for relevance detection and the extension of adaptive mechanisms in large-scale repositories. The practical significance of the study lies in the possibility of improving the efficiency of adaptive learning systems, optimizing the organization of multimedia databases, and strengthening software engineering solutions for the development of intelligent educational platforms. Therefore, the proposed approach can be used as a formal basis for building adaptive multimedia systems capable of handling big data while preserving semantic consistency and pedagogical relevance.