Neural Network Approach for Semantic Coding of Words

This article examines and analyzes the use of the word2vec method for solving semantic coding problems. The task of semantic coding has acquired particular importance with the development of search system. The relevance of such technologies is associated primarily with the ability to search in large-volume databases. Based on the obtained practical results, a prototype of a search system based on the use of selected semantic information for the implementation of relevant search in the database of documents has been developed. Proposed two main scenarios for the implementation of such a search. The training set has been prepared on the basis of documents in the English version of Wikipedia, which includes more than 100,000 original articles. The resulting set was used in the experimental part of the work to test the effectiveness of the developed prototype search system.