Optimization of Marketing Strategies Based on Social Graph Analysis

This thesis focuses on the possibility of obtaining additional value for the company, which has data on social connections of users. Attention in this work has been paid to the sphere of digital advertising, namely, conducting advertising campaigns on the Internet. The advertising campaign can be optimized by obtaining a relevant target audience, through the use of data on social connections. Knowing the interest of users in some area, it is possible to get a segment of users who will have a similar interest base on the data about their communications. That is, the company can expand the audience that is more inclined to purchase goods at the expense of their data. In this paper data from two different sources have been analyzed. In the first case, the data obtained through the API on the author's communication circle in Vkontakte social network has been analyzed. Various mechanisms have been used to determine the importance of nodes in the graph. It was also determined that there is no correlation between the algorithms. This allowed to independently choose the algorithm for further analysis that proved to be the most suitable on the first graph. In the second case, the graph based on user communications and the duration of their communications has been analyzed. Thus, each edge of the graph was given its weight. In the resulting graph there were users who were interested in the field of pharmacology. Two algorithms have been used to determine the relevant audience: PageRank and HITS, which have been optimized for the conditions of the problem of this thesis. To check the quality of the algorithms, a real test campaign has been carried out, which showed that users obtained using algorithms have better performance than the general audience (users aged 25-35 years). It should be concluded that the constructed algorithms can be applied in practice. In conclusion the additional research options are discussed to provide a future outlook.