Методи глибинного навчання та обробки природномовних текстів в задачах психологічного профілювання людей

Abstract

This thesis is related to the field of natural language processing (NLP) and psychology, namely personality classification by the Myers-Briggs type indicator (MBTI). The main task that has been resolved during the research is methods for text classification for such subject area comparison and data preprocessing. Several common approaches for text classification and dataset balancing have been used to retrieve and compare the metrics of the models. The main problem was a totally unbalanced dataset that had provoked a decrease in metrics in terms of quality. The results of the research described here show the effectiveness of using deep neural network architectures for the task of text classification in English inside the presented subject area