Methods for mental operations detection in scientific publications

The paper presents results of a study of methods for a mental structure detection in scientific text. Initially, we consider the concepts of mental operations and mental structure of a scientific text. The reasons for choosing the taxonomy for mental operations used in our approach are also presented. We set the problem of mental operations detection in texts and propose the methods for solving it. An empirical study on an annotated dataset and on a corpus of more than 7k scientific articles showed the applicability of the proposed methods for detection a mental structure of a scientific text. Informative features of mental structures of texts from different research areas were also detected.