Extension of a standard balanced linguistic corpus built according to spaCy rules by connotative characteristics

Abstract

The aim of the research is to develop the technology for automatically determining the sentiment of a text based on the existing author’s software package. The scientific novelty lies in the fact that the work proposes a structural and functional model of a fully automated process for assessing the sentiment of a text in conjunction with an analysis of its morphological characteristics; the technical terms “connotative amplitude” and “connotative density” are also introduced for the first time. The study built a database model that accommodates connotative numeric parameters; further, the program code for the “add-on” for the database generator has been written, which allows one to supplement the standard database with these parameters; finally, the technology was tested on the material of three novels by F. Kafka (“Castle”, “The Trial” and “America”) and two novels by E. M. Remarque (“All Quiet on the Western Front” and “Flotsam”) in the German language. As a result, it is proven that the “add-on” is a high-quality software product that does not cause technical failures and is capable of providing researchers with a whole set of connotative data for subsequent comprehensive interpretation of the text, on condition that the input tone dictionary is of high quality.