TSBuilder 2.0: Improving the Identification Accuracy Due to Synonymy

The paper reviews a project on the automation of term system construction. TSBuilder (Term System Builder) was developed in 2014 as a multilayer Rosenblatt's perceptron for supervised machine learning, namely 1-3 word terms identification in natural language texts and their rigid categorization. In the next version we expanded the range of categorization parameters, removed the term length restriction, used convolution for a continuous sequence of terms, and analyzed the probability of term's falling into different categories.In the current release, we added the use of synonymic matching for single- and multiple-word terms. This allows us to replenish the database with new words close in meaning to the existing ones but having dropped out of our radar for not being cognit to the ones included into the database. Algorithms for translating natural language words into a vector form were used to estimate the degree of synonymy.