This paper describes the methods and techniques used to effectively extract knowledge from large volumes of heterogeneous data. Also, methods to structure the raw data by the automatic classification using ontology are discussed. In the first part of the article the basic technologies to realize the Semantic WEB are described. Much attention is paid to the ontology, as the major concepts that structure information on a very high level. The second part examines AVT-DTL algorithm proposed by Jun

more »

... hang which allows one to automatically create classifiers according to the available raw, potentially incomplete data. The considered algorithm uses a new concept of floating levels of ontology; the results of the tests show that it outperforms the best existing algorithms for creating classifiers.