Collecting the Database for the Neural Network Deep Learning Implementation

Abstract

The paper reviews the work done by the research group in the field of text mining for automating term system construction, i.e., term identification and categorization. The automation is achieved via the usage of binary search trees, a multilayer version of the Rosenblatt’s perceptron. The algorithm comprises elements of supervised machine learning with given classes. The paper provides detailed description of the methods and digital resources (online dictionaries, taggers, and corpora) used for manual collection of the database for the subsequent neural network learning. Finally, we present our perspectives on the developed software modification with the transition to deep learning.