Abstract: Social networks provide an opportunity to form communities of people that share their interests on a regular basis (circles of fans of different music, books, kinds of sports, etc.). Every community manifests these interests creating lots of linguistic data to attract new followers to certain pages and support existing clusters of users. In the present article, we suggest a model of retrieving such pages that attract users with similar interests, from a large collection of pages. We test our model on three types of pages manually retrieved from the social network Vkontakte and classified as interesting for a. football fans, b. vegetarians, c. historical reenactors. We use such machine learning classifiers as Naive Bayes, SVM, Logistic Regression, Decision Trees to compare their performance with the performance of our system. It appears that the mentioned classifiers can hardly retrieve (i.e. single out) pages with a particular interest that form a small collection of 30 samples from a collection as large as 4,090 samples. In particular, our system exceeds their best result (F1-score=0.65) and achieves F1-score of 0.72.

Abstract: Models of information spread in social networks are widely used to explore the drivers of content contagion and to predict the effect of new information messages. Most of the existing models (aggregated as SIR-like or network-based as independent cascades) use the assumption of homogeneity of an audience. However, to make a model plausible for a description of real-world processes and to measure the accumulated impact of information on individuals, one needs to personalize the characteristics of users as well as sources of information. In this paper, we propose an approach to data-driven simulation of information spread in social networks which combines a set of different models in a unified framework. It includes a model of a user (including sub-models of reaction and daily activity), a model of message generation by information source and a model of message transfer within a user network. The parameters of models (e.g. for different types of agents) are identified by data from the largest Russian social network vk.com. For this study, we collected the network of users associated with charity community (~33.7 million nodes). To tackle with huge size of networks, we implement-ed parallel version of modeling framework and tested it on the Lomonosov supercomputer. We identify key parameters of models that may be tuned to re-produce observable behavior and show that our approach allows to simulate aggregated dynamics of reactions to a series of posts as a combination of individual responses.

Abstract: The network structure of communities in social media significantly affects diffusion processes which implement positive or negative information influence on social media users. Some of the thematic communities in online social networks may provide illegal services or information in them may cause undesired psychological effects; moreover, the topology of such communities and behavior of their members are influenced by a thematic. Nevertheless, recent research does not contain enough detail about the particularities of thematic communities formation, or about the topological properties of underlying friendship networks. To address this gap, in this study we analyze structure of communities of different types, namely, carders, commercial sex workers, substance sellers and users, people with radical political views, and compare them to the 'normal' communities (without a single narrow focus). We discovered that in contrast to ordinary communities which have positive assortativity (as expected for social networks), specific thematical communities are significantly disassortative. Types of anomalous communities also differ not only in content but in structure. The most specific are the communities of radicalized individuals: it was shown that they have the highest connectivity and the larger part of nodes within a friendship graph.

Abstract: Complex networks can be used for modeling street meshes and urban agglomerates. With such a model, many aspects of a city can be investigated to promote a better quality of life to its citizens. Along these lines, this paper proposes a set of distance-based pattern-discovery algorithmic instruments to improve urban structures modeled as complex networks, detecting nodes that lack access from/to points of interest in a given city. Furthermore, we introduce a greedy algorithm that is able to recommend improvements to the structure of a city by suggesting where points of interest are to be placed. We contribute to a thorough process to deal with complex networks, including mathematical modeling and algorithmic innovation. The set of our contributions introduces a systematic manner to treat a recurrent problem of broad interest in cities.