An algorithm creating for predicting the inaccurate information presence in social networks in Russian language

The development of user-to-user communication methods, such as social media, has resulted in the amount of inaccurate information reaching record levels. This problem affects not only regular users of social media, but also the media, which may refer to such messages as a source of information. The spread of false information leads to both financial and life-threatening problems. It is almost impossible to trace these messages manually and therefore it is required to create an algorithm that can perform this process automatically. The purpose of this paper is to try to create such an algorithm for the Russian language using machine learning methods. The data on which the models are based is a sample of data that has undergone the process of manual annotation. The sample has undergone the process of preparation and balancing. From this sample, 29 attributes were obtained which can be divided into 3 categories: user, text and distribution. These attributes and were applied to obtain classification models that are able to predict with sufficiently high probability. The result of this work is an algorithm for predicting the presence of inaccurate information in a social network post.