PROBLEM OF CLASSIFICATION OF SEMANTIC KERNELS OF WEB RESOURCE

Abstract

The article presents a new theoretical basis for solving the problem of situational management of semantic cores identified on the basis of WEB content. Such a task arises within the framework of a new phenomenon called virtual promotion. Its essence lies in the fact that a real product can exist in two realities: online and offline. According to marketing theory, the lifetime in two realities is the same. However, in the online mode, the goods exist independently and in accordance with the laws of the use of Internet technologies. Therefore, based on the concept of a marketing channel, it was proposed to consider a message in such a channel as a semantic core. The core is a specially selected set of keywords that briefly describe the product and the corresponding need. It has been proposed that each need forms a so-called class of need. Therefore, the product description will either belong to this class or not. In addition, a product can be described by a different set of keywords, which means that different descriptions of the same product or several products, if there are any for sale in the enterprise, will fall into the demand class. As a result, in this work, it was proposed to consider the center of this class as the so-called K-candidate. It is the K-applicant that will be the semantic core that will be considered at the current iteration of the situational management process. In addition, in order to move from one situation to another, in other words, from one core to another, it is required to have such an alternative core. It can be safely taken either from the neighborhood of the need class center (K-applicant), or the center of another class (another K-applicant), if the product can cover several needs of a potential buyer. Then the actual task is to classify the classes of needs based on the text corpus in HTML format. Having a text corpus at the first stage, the task of synthesizing semantic cores is realized, and then the classification task itself. This article proposes the formulation of the classification problem, taking into account the features that the Internet technologies contribute to search engine optimization. In particular, it is proposed to use four metrics from the category of WEB statistics. And then it is proposed to use the clustering method to identify classes of needs, taking into account the fact that the K-applicant is presented as a semantic network or as a graph.