The paper is dedicated to solving the problem of optimal text classification in the area of automated detection of typology of texts. In conventional approaches to topicality-based text classification (including topic modeling), the number of clusters is to be set up by the scholar, and the optimal number of clusters, as well as the quality of the model that designates proximity of texts to each other, remain unresolved questions. We propose a novel approach to the automated definition of the optimal number of clusters that also incorporates an assessment of word proximity of texts, combined with text encoding model that is based on the system of sentence embeddings. Our approach combines Universal Sentence Encoder (USE) data pre-processing, agglomerative hierarchical clustering by Ward’s method, and the Markov stopping moment for optimal clustering. The preferred number of clusters is determined based on the “e-2” hypothesis. We set up an experiment on two datasets of real-world labeled data: News20 and BBC. The proposed model is tested against more traditional text representation methods, like bag-of-words and word2vec, to show that it provides a much better-resulting quality than the baseline DBSCAN and OPTICS models with different encoding methods. We use three quality metrics to demonstrate that clustering quality does not drop when the number of clusters grows. Thus, we get close to the convergence of text clustering and text classification.

The leukocyte formula is the percentage of different groups of white blood cells. According to morphological features, three subpopulations can be distinguished among leukocytes: lymphocytes, monocytes and granulocytes. Granulocytes are divided into neutrophilic, eosinophilic, and basophilic cells. Automatic typologization of white blood cells is an unsolved problem, since at present, during cytometric research, the counting of the number of cells in various subpopulations of leukocytes is actually done manually, which in turn causes the subjectivity of the experiment and large values of errors in calculations. To solve this problem, attempts have been made repeatedly to use cluster analysis methods. In computational experiments, it was shown that the use of standard algorithms, such as the agglomerative methods, EM algorithm, DBSCAN, etc., does not allow to obtain the desired results. In recent years, a large number of research papers have been published describing specialized clustering algorithms for detecting and determining populations of white blood cells, some of them have found practical application, but the problems associated with the presence of a large amount of noise and different data density distribution during leukocyte clustering by flow cytometry methods remain relevant. The article considers an approach to constructing a strategy for automatic allocation of the main leukocyte subpopulations using a modified agglomerative centroid clustering method and discusses the results of computational experiments. The results of calculating the proportion of lymphocytes are compared “manually” and automatically using a modified centroid algorithm.