Authorship Attribution of Social Media and Literary Russian-Language Texts Using Machine Learning Methods and Feature Selection

Abstract

Authorship attribution is one of the important fields of natural language processing (NLP). Its popularity is due to the relevance of implementing solutions for information security, as well as copyright protection, various linguistic studies, in particular, researches of social networks. The article is a continuation of the series of studies aimed at the identification of the Russian-language text’s author and reducing the required text volume. The focus of the study was aimed at the attribution of textual data created as a product of human online activity. The effectiveness of the models was evaluated on the two Russian-language datasets: literary texts and short comments from users of social networks. Classical machine learning (ML) algorithms, popular neural networks (NN) architectures, and their hybrids, including convolutional neural network (CNN), networks with long short-term memory (LSTM), Bidirectional Encoder Representations from Transformers (BERT), and fastText, that have not been used in previous studies, were applied to solve the problem. A particular experiment was devoted to the selection of informative features using genetic algorithms (GA) and evaluation of the classifier trained on the optimal feature space. Using fastText or a combination of support vector machine (SVM) with GA reduced the time costs by half in comparison with deep NNs with comparable accuracy. The average accuracy for literary texts was 80.4% using SVM combined with GA, 82.3% using deep NNs, and 82.1% using fastText. For social media comments, results were 66.3%, 73.2%, and 68.1%, respectively.

[Retracted] Analysis of Sino‐Russian Media Cooperation and the Construction of International Online Public Opinion Discourse under the Dual Influence of Ecological and Online Environments

“…Some scholars emphasize that the core of public opinion guidance lies in the competition for the right to speak, think that the main elements of the right to speak are the right to speak, the right to spread, and the right to guide, and put forward a new path of public opinion guidance. Some scholars pointed out that news and public opinion are an important carrier for a political party to master the right to speak [ 9 ]. The discourse power of news public opinion should obey certain political needs and guide the trend of news public opinion.…”

Privacy-Preserving and Explainable AI in Industrial Applications

“…• saliency maps: elements in the input that have the largest influence in the prediction are identified (e.g., LIME); • feature attribution: attributing the classification to a small number of numeric/semantic features [47,48]; • metric learning [49]: mapping out data structures by deriving a metric from a classifier (explicit Siamese networks are very popular); • activation maximization: methods that are based on GAN.…”

Author Identification with Machine Learning Algorithms

“…An important part of the literature consists of studies on English language [4,5,6,7,8]. There are also many studies done in many different languages including Japanese [9], Mongolian [10], Persian [11], Albanian [12], Indian [13,14], Brazilian [15], Russian [16,17], German [18], and Arabic [19]. When the existing studies were examined, it was seen that different types of data sets were used for author identification tasks.…”

“…Some studies have been carried out on newspaper articles [4,15,18,19], while others were carried out on poems [13], novels [11,12,16], email content [20], song lyrics [21], source codes [22], or tweets, blog posts, and forums [8,9,23]. In some cases, different types of data sources were combined or compared [17,25] Early studies in author identification focused on different stylometric techniques. These techniques are based on identification of style markers including lexical and character features or syntactic and semantic features that quantify writing style [9,26].…”