Russian Text Vectorization: An Approach Based on SRSTI Classifier

Abstract

This paper presents an approach to Russian text vectorization based on SRSTI classifier. Our approach is based on using SRSTI categories as vector space dimensions. The categories are defined by lists of keywords. We explain our choice of SRSTI as a basis for vector space. We describe the keywords selection process, as well as vector calculation and comparison algorithm. We apply developed algorithm to marked-up SRSTI texts and user social profiles. We also suggest approaches to vector space improvement and evaluate them.