Evaluation of the Coherence of Polish Texts Using Neural Network Models

Coherence evaluation of texts falls into a category natural language processing tasks. The texts’ coherence implies the estimation their semantic and logical integrity; such feature text can be utilized during solving multidisciplinary tasks (SEO analysis, medicine area, detection fake texts, etc.). In this paper, different state-of-the-art methods based on machine learning models have been analyzed. investigation effectiveness for Polish has performed. impact text’s features output value analyzed using approaches similarity graph. Two neural networks LSTM layers pre-trained BERT model correspondingly designed trained input texts. results obtained may indicate that both lexical components should taken account documents; moreover, it is advisable to analyze corresponding documents in sentence-by-sentence manner taking word order. According retrieved accuracy proposed networks, concluded suggested used order solve typical corpus.