Although conflict representation in media has been widely studied, few attempts have been made to perform large-scale comparisons of agendas in the media of conflicting parties, especially for armed country-level confrontations. In this paper, we introduce quantitative evidence of agenda divergence between the media of conflicting parties in the course of the Ukrainian crisis 2013–2014. Using 45,000 messages from the online newsfeeds of a Russian and a Ukrainian TV channels, we perform topic modeling coupled with qualitative analysis to reveal crisis-related topics, assess their salience and map evolution of attention of both channels to each of those topics. We find that the two channels produce fundamentally different agenda sequences. Based on the Ukrainian case, we offer a typology of conflict media coverage stages.
This paper investigates to what extent activity of a social movement on a social networking site is related to participation in offline collective action. Through this research, we seek to contribute to a broader theory of effective communicative structures of social movements. We use the data of roughly 12,000 individuals from 17 online groups representing the branches of the ‘Observers for Fair Elections’ movement in 17 districts of St. Petersburg, Russia, and compare their online properties to real offline participation of movement members in elections in the role of electoral observers. We find that while prediction of individual offline participation with this online data is of limited power, association between district participation rates and online group features is very strong. Large, more inclusive and evenly connected networks, where people are engaged in high-threshold online activities, produce more offline participants; weak individual-level prediction, combined with strong group-level prediction, suggests either the presence of the ‘network effect’ or of third factors – such as prior contentious experience or the effect of leaders.
In the framework of this paper we apply multifractal formalism to the analysis of statistical behaviour of topic models under variation of the number of topics. Fractal analysis of topic models allows to show that self-similar fractal clusters exist in large textual collections. We provide numerical results for 3 topic models (PLSA, ARTM, LDA Gibbs sampling) on 2 datasets, namely, on an English-language dataset and on a Russian-language dataset. We demonstrate that forming of clusters occurs precisely in the transition regions. Linear regions do not lead to changes in fractals, therefore, it is sufficient to find transition regions for the study of textual collections. Accordingly, the problem of the analysing the evolution of topic models can be reduced to the problem of searching transition regions in topic models.
This study proposes to minimize Rényi and Tsallis entropies for finding the optimal number of topics T in topic modeling (TM). A promising tool to obtain knowledge about large text collections, TM is a method whose properties are underresearched; in particular, parameter optimization in such models has been hindered by the use of monotonous quality functions with no clear thresholds. In this research, topic models obtained from large text collections are viewed as nonequilibrium complex systems where the number of topics is regarded as an equivalent of temperature. This allows calculating free energy of such systems—a value through which both Rényi and Tsallis entropies are easily expressed. Numerical experiments with four TM algorithms and two text collections show that both entropies as functions of the number of topics yield clear minima in the middle area of the range of T. On the marked-up dataset the minima of three algorithms correspond to the value of T detected by humans. It is concluded that Tsallis and especially Rényi entropy can be used for T optimization instead of Shannon entropy that decreases even when T becomes obviously excessive. Additionally, some algorithms are found to be better suited for revealing local entropy minima. Finally, we test whether the overall content of all topics taken together is resistant to the change of T and find out that this dependence has a quasi-periodic structure which demands further research.
A one-parameter family of Mackey-Glass type differential delay equations is considered. The existence of a homoclinic solution for suitable parameter value is proved. As a consequence, one obtains stable periodic solutions for nearby parameter values. An example of a nonlinear functions is given, for which all sufficient conditions of our theoretical results can be verified numerically. Numerically computed solutions are shown.
Online petitions are usually regarded as one of the most popular channels to involve citizens in the political process. In our paper we have analyzed texts and voting data (pro and against) from 9705 e-petitions submitted from 2013 until 2017 at Russian Public Initiative project. Analysis of dynamics showed stabilization of interest to this resource (emergence of a new authors, growth of “strong” petitions etc.). Studying success factors of electronic petitions at the Russian public initiative project we found out that the topic and lexical information are significant factors, as well as the level of petitions.
The paper reveals the topic structure of ethnic discussions in the Russian-speaking social media and explores how these topics are related to the post-Soviet ethnic groups. Analyzed more than 2.6 million texts from Russian-speak- ing social media published for two-year period from 2014 to 2015 and contained at least one of the post-Soviet ethnonyms, we conclude that ethnic discussions in these media are full of socially significant and potentially problematic topics (15 topics out of 97 can be regarded as problematic comparing to the 4 out of 150 topics on random sample from VK.com). The most salient topics are the topics about Ukraine-Russia relations over the recent conflict between two countries. We also found the racial bias in criminal topic towards peoples of the North Caucasus which are often mentioned in the context of crimes and terrorism.
Out-group bias in the context of race and ethnicity has been widely studied. However, little research has been done to study this phenomenon online. In this paper we explore how ethnicity and gender of Russian social media users affects their attitude toward other ethnic groups. Out results show that ethnicity of social media users plays a significant role in their attitude towards ethnic groups. On the average social media users tend to experience more positive attitude to the ethnic groups they belong to, but there are some exceptions — Russians and Tatars. Gender also influence the attitude insofar as men expressed stronger negative emotions toward foreign peoples.
This book constitutes the proceedings of the 5th International Conference on Analysis of Images, Social Networks and Texts, AIST 2016, held in Yekaterinburg, Russia, in April 2016. The 23 full papers, 7 short papers, and 3 industrial papers were carefully reviewed and selected from 142 submissions. The papers are organized in topical sections on machine learning and data analysis; social networks; natural language processing; analysis of images and video.
The advent of personalized medicine and wide-scale drug tests has led to the development of methods intended to automatically mine and extract information regarding drug reactions from user reviews. For medical purposes, it is often important to know demographic information on the authors of these reviews; however, existing studies usually either presuppose that this information is available or disregard the issue. We study automatic mining of demographic information from user-generated texts, comparing modern natural language processing techniques, including extensions of topic models and deep neural networks, for this problem on datasets mined from health-related web sites.
The ability of social media to rapidly disseminate judgements on ethnicity and to influence offline ethnic relations creates demand for the methods of automatic monitoring of ethnicity-related online content. In this study we seek to measure the overall volume of ethnicity-related discussion in the Russian-language social media and to develop an approach that would automatically detect various aspects of attitudes to those ethnic groups. We develop a comprehensive list of ethnonyms and related bigrams that embrace 97 Post-Soviet ethnic groups and obtain all messages containing one of those words from a two-year period from all Russian-language social media (N=2,660,222 texts). We hand-code 7,181 messages where rare ethnicities are over-represented and train a number of classifiers to recognize different aspects of authors’ attitudes and other text features. After calculating a number of standard quality metrics, we find that we reach good quality in detecting intergroup conflict, positive intergroup contact, and overall negative and positive sentiment. Relevance to the topic of ethnicity and general attitude to an ethnic group are least well predicted, while some aspects such as calls for violence against an ethnic group are not sufficiently present in the data to be predicted.
An important role of digital inequality for hindering the development of civil society is being increasingly acknowledged. Simultaneously, differences in availability and the practices of use of social network sites (SNS) may be considered as major manifestations of such digital divide. While SNS are in principle highly convenient spaces for public discussion, lack of access or domination by socially insignificant small talk may indicate underdevelopment of the public sphere. At the same time, agenda differences between regions may signal about local problems. In this study we seek to find out whether regional digital divide exists in such a large country as Russia. We start from a theory of uneven modernization of Russia and use the data from its most popular SNS “VK.com” as a proxy for measuring digital inequality. By analyzing user activity data from a sample of 77,000 users and texts from a carefully selected subsample of 36,000 users we conclude that regional level explains an extremely small share of variance in the overall variation of behavioral user data. A notable exception is attention to the topics of Islam and Ukraine. However, our data reveal that historically geographical penetration of “VK.com” proceeded from the regions considered the most modernized to those considered the most traditional. This finding supports the theory of uneven modernization, but it also shows that digital inequality is subject to change with time.
This book constitutes the refereed proceedings of the First International Conference on Digital Transformation and Global Society, DTGS 2017, held in St. Petersburg, Russia, in June 2017.
The 34 revised full papers and three revised short papers presented were carefully reviewed and selected from 134 submissions. The papers are organized in topical sections on eSociety: social media analysis; eSociety: ICTs in education and science; eSociety: legal, security and usability issues; ePolity: electronic governance and electronic participation; ePolity: politics of cyberspace; eCity: urban planning and smart cities; eHealth: ICTs in public health management; eEconomy and eFinance: finance and knowledge management.
Deep generative adversarial networks (GANs) are the emerging technology in drug discovery and biomarker development. In our recent work, we demonstrated a proof-of-concept of implementing deep generative adversarial autoencoder (AAE) to identify new molecular fingerprints with predefined anticancer properties. Another popular generative model is the variational autoencoder (VAE), which is based on deep neural architectures. In this work, we developed an advanced AAE model for molecular feature extraction problems, and demonstrated its advantages compared to VAE in terms of (a) adjustability in generating molecular fingerprints; (b) capacity of processing very large molecular data sets; and (c) efficiency in unsupervised pretraining for regression model. Our results suggest that the proposed AAE model significantly enhances the capacity and efficiency of development of the new molecules with specific anticancer properties using the deep generative models.