The highest dynamics of the processes of online extremism bring more importance to the development of methodology and software tools capable to produce real-time tracking of the dissemination of information, including social networks, to analyze posts meaning and promise and to build predictive models. The article presents the technique of designing a smart linguistic models that are able to take into account the context and allows flexible adaptation to the subject area for natural language processing in the social networks within the field of real-time warning cyber-threats based on Data Mining. We describe the basic disadvantages of using a simple feature engineering and the bag of words method for the purposes of text classification. We show programming interface and features of the generalized framework in which our technique is applied, and also show how this framework can be used to meet the challenges of business and government during the process of collection and analysis of publications on the Internet.

Keywords: natural language processing, linguistic models, machine learning, feature engineering, text mining framework, text classification, language models constructor, morphology analysis

We analyze social network graph to identify popularity distribution between ukrainian presidential candidates during elections period. In the first part of the paper we descibe 2 graph layouts: candidates in-degree and bloggers in-degree. It is stated, that with the help of such data mining one can answer community-centric questions: community clusters, that mention same candidates, structure of such mentions, particular bloggers, that mention particular candidates.

Keywords: social network graph, data mining, big data, opinions graph, presidential candidates, graph theory, social network analysis

A detailed description of the platform is provided. We describe key features of the university's indices database including secureness, structure flexibility, access abilities. We demonstrate how user interfaces can be used to fill in database. Also we suggest platform's filling algorithm.

Keywords: information platform, effectiveness indices, relational database

To analyze social network graph effectively one needs a web-based tool, to parse network data automatically. In the first part of the paper we descibe a structre of the data we gain, including who comments whom information. Then we explain usage of Drupal CMF with Feeds module to parse network data and save it to database. The data obtained consist of name of the comment author, date posted, journal author, comment text, title of the comment. Prepared graph can be exported for further use.

Keywords: social network graph, web-page parser, drupal cmf, feeds module, comments importer, comments graph

  Nowadays, a dramatic increase in the influence of events of the blogosphere onto the world's political and social processes is observed. We in Northern - Caucasian Scientific Center of High School have developed techniques of social media monitoring and modeling of the structure of the thematic blogosphere segment. During monitoring process the following attributes are under analysis: sense of text, types, complexity and set of features of objects, types of relations between the objects themselves, etc. - to obtain complete range of blogosphere users’ opinions on a particular theme. The structure of the politically active segment of the social network is modeled with the help of graph, using appropriate mathematical methods of analysis. With the help of the technique been developed research of the social media communications was carried out in the context of the State Duma elections.

Keywords: blogosphere, modeling, monitoring, election campaigning, social media, social networks