Method for determination of personal opinion leader based on social network analysis

<p align="center">Abstract</p><p align="center">For Stefan Fateev</p><p align="center">Master Thesis:</p><p align="center">&ldquo;Method for determination of personal opinion leader based on social network analysis &rdquo;</p><p align="center">&nbsp;</p><p>The paper examines a problem of identifying an effective communication channel&nbsp; in social networks. In particular, using formal methods and metrics of social network analysis (such as Betweenness, Closeness etc.) for finding opinion leaders in egocentric graphs has been shown. This work focused on the trust phenomenon, which exists between target and opinion leader. For this purpose a prediction model of trust relations between target and its inner circle has been built. The model based on analysis of selected characteristics of its bearers using the structure of the social network. A data for this research was mined from largest Russian social network, called vk.com. Eighty real egocentric graph were constructed and analyzed. Information block related to the trust phenomenon was obtained from internet questioner. In the issue data pull consists 395 observations. Binary logistic regression is selected as a method for building a model of trust relations. Researchers conclude that Betweenness plays a key role for identifying trust relations.</p><p>The study confirmed the existence of a direct relationship between trust in the person and the influence of its recommendations on the decision to purchase FMCG goods, and there is a direct relationship between trust in the person and willingness to listen to his advice on unknown services (goods). These results give reason to use the resulting model in WOM-marketing.</p><p>Substantial work has been carried out from summarizing existing ways of finding opinion leaders and social network analysis, to developing its own model. Using formal methods of social network analysis to assay the interaction of participants of social networks has significant theoretical interest, and this work may serve as a starting point for developing further applications.</p><p>Designed approach can be used to achieve a variety of marketing purposes. In particular, &nbsp;a number of effective mechanisms for use in WOM-marketing, especially in recommendation services is suggested.</p><p>In terms of methodology, the restriction of a small sample of participants in experiments, make it not possible to achieve statistical significant results on some hypotheses. Researcher gives a number of recommendations and suggestions for further research.</p>