The problem of automatic understanding of full text documents in information retrieval

New opportunities for retrieval and automatic processing of full text documents that arise in the use of modern intelligent systems are discussed. An overview of major limitations of the classical scheme of descriptive information retrieval is presented, and a direction of its modernization is proposed. This direction leans on computer reconstruction of concepts on the basis of documents that discuss them. The key role of an intelligent system of linguistic analysis and machine learning in this process is characterized. An overview of sample computer experiments in thematic classifications of natural language texts is presented. The perspectives of application of the proposed approach to a number of research and industrial projects are outlined.