Prediction of Product Adoption in Social Networks Using the Network Value of Users

Ilya Verenich
In this work we study the adoption of a product in a social network with the purpose of determining the set of users to target during a marketing campaign to maximize the campaign return. As a baseline, we use a model to estimate users' propensity to adopt the product. The model is trained and evaluated on temporally split data and shows a significant lift over random guessing. We also find the strong evidence of the peer pressure in our network. To utilize the network value of users, we infer interpersonal influence with the notion of temporally correlated adoptions. Then we design a model to determine influential network users, who, given that they adopt the product, will trigger subsequent adoptions among their friends. Finally, we introduce the concept of a users' utility that combines users' propensity to adopt the product with their potential influence on their friends. On a simulated marketing campaign we show that targeting a fixed number of high-utility users results in more adoptions, than targeting either highly influential users or users with high propensity to adopt, which confirms the practical value of our complementary approach.
Graduation Thesis language
Graduation Thesis type
Master - Software Engineering
Riivo Kikas, Marlon Dumas
Defence year