The ability to automatically determine the age audience of a novel provides many opportunities for the development of information retrieval tools. Firstly, developers of book recommendation systems and electronic libraries may be interested in filtering texts by the age of the most likely readers. Further, parents may want to select literature for children. Finally, it will be useful for writers and publishers to determine which features influence whether the texts are suitable for children. In this article, we compare the empirical effectiveness of various types of linguistic features for the task of age-based classification of fiction texts. For this purpose, we collected a text corpus of book previews labeled with one of two categories -children's or adult. We evaluated the following types of features: readability indices, sentiment, lexical, grammatical and general features, and publishing attributes. The results obtained show that the features describing the text at the document level can significantly increase the quality of machine learning models. Keywords: Text classification • Fiction • Corpus • Age audience • Content rating • Text difficulty • RuBERT • Neural network • Natural language processing • Machine learning Supported by the grant of the President of the Russian Federation no. MK-637.2020.9.

No abstract

In recent years, modeling of human actions and activity patterns for recognition or detection of the special situation has attracted a significant research interest. We present our approach for abnormal human action recognition as a sequence of intermediate states. We propose to decompose each action into a sequence of discrete intermediate states and to present state transitions as a stochastic process. Each state is described with the joint locations of a human skeleton. Actions are described with Hidden Markov Model based on the found states and its interconnections. As a result, we combine our stochastic model of human actions with intermediate states described via skeleton joints. Convolutional Neural Network is employed to learn skeleton features for intermediate state recognition. Viterbi algorithm is employed to find model parameters. We implemented proposed methods in a framework for human abnormal action recognition and tested our approach on two samples: MPII Human Pose Dataset and exam footages.

No abstract