Tasks of author attribution of text in the aspect of gender (on interdisciplinary interaction of linguistics and computer science)

The paper discusses the interdisciplinary interaction of linguistics and computer science in connection with solving the tasks of author attribution of the text in the aspect of gender. The problem areas of theory and practice of authorship expertise are defined. Since the problem of attribution of the author of the text is now addressed and resolved mainly on the material of works of art, we should solve the question of how these techniques will work if to use them on text material of other styles and discourses. There is a need to develop the problem of combining qualitative and quantitative methods of analysis and definition of the type of language units, statistical analysis of which is sufficient to identify the text. The paper points out common problem areas of gender linguistics and authorship expert works that need solutions. The authors emphasize the importance of involving findings of authorship expertise experiences in the paradigm of gender research, especially on unreflected formal attributes of the speech itself, as these are great signs of the opposition of male and female speech behaviour. There is a need in quantitative verification of features of "typical male" and "typical female'' speeches selected by qualitative methods. Authorship expertise requires active use of latest gender researches in the analysis. The solution of the complex tasks stated in the paper on theoretical and practical problems will be carried out in an interdisciplinary project performed by a team of authors from the Department of Complex Information Security of Computer Systems of Tomsk State University of Control Systems and Radioelectronics and the Department of General, Slavic-Russian Linguistics and Classical Philology of Tomsk State University. A research program was designed aimed at identifying the impact on the effectiveness of the applied methods of author attribution settings of 1) areas of communication, 2) length of the fragments of communication, and 3) ratio of detected asemantic and semantic units of text structures. Solution of the problems of authorship expertise requires a special empirical basis of research. The objectives of the first phase of the project are: 1) creation of a corpus of text of various discourses and marking it in accordance with the tasks of authorship expertise (in the first phase to form a corpus of short electronic messages is planned and its linguistic marking), 2) expansion of the list of features used to identify gender and studies on determining their information capacity for the problem, which will lead to the expansion of the functionality of the software complex "Avtoroved", 3) study of meta-classifiers in order to improve the accuracy of identification of the author of the text, 4) justification of the results by mathematical and contemporary linguistic methods, 5) synthesis of the previously obtained models and techniques to identify the author of the text.

Keywords

Authors

References