A writer identification is one of the fundamental problems of natural language processing. An establishment of a document author plays an important role in linguistic, historical and criminalistic researches. A great number of texts has become available with the development of the Internet, which made it virtually impossible to manually classify texts by authors. Thus, the task automatically finding authorship is currently very topical. 	Currently, there are several commercial solutions for the determination of the author of the text, but they all have some disadvantages associated with the accuracy and robustness of used methods. Modern algorithms of a writer identification based on statistical analysis and conclusion of the specific feature set derived from the text. One of the most popular are machine learning techniques such as neural networks, support vector machines, hidden Markov chains, decision trees. For their stable operation needed training data, the volume of which is directly related to the accuracy and stability of the algorithms listed above. It is also an important factor for practical applications is the speed of the method, as is often the system should work in real time. 	In her qualifying work Polina studied the using of the method of feature extraction from the text, based on an assessment of the distribution of frequency of successive combinations of letters (n-grams). This method has shown very prominent results in a variety of problems texts processing, such as the problem of determining the part of speech. Polina in her work has adapted this approach to determine the authorship of the text. Also, in work was analysed the using of different clustering methods with the features obtained by n-grams. Such algorithms were considered as K-Means, Global K-Means, PAM. Analysis of their work evaluated by a variety of clustering quality metrics: Rand Index, NMI, F-Measure, Purity measure. This approach allowed to obtain diverse and balanced assessment of the results. The great advantage of the provided in this work clustering methods is that they have shown consistent results in the problem of determining the authorship of the text, as well as, by virtue of its simplicity, demonstrated a high processing speed. In this paper, on the basis of the study were made conclusions about the terms of the applicability of the method based on the distribution of frequencies of combinations of letters, as well as its limitations in defining the author of the text. The results can serve as a basis for further research in this area, and find practical application in the field of natural language processing. 	As part of the qualification work by Polina has been presented the software implementation of the described method in Python language with using optimized libraries. This system allows to adjust the method of architecture to work with Russian, English and German. 	As a result, I can conclude that this work deserves the excellent mark.