Mitigating Uncertainty In Ai-Driven Marketing: Operational Learning Pipeline

AI and neural networks are well known for their effectiveness in working with large amounts of data under uncertainty. They play a key role in most recommendation systems. However, recommendation systems are designed to work with existing users already known to the system. Equally or even more important is working with new users to attract them to the system when no information about them is yet available. Despite the problem's rel-evance, only a few studies have addressed the issue of user acquisition us-ing neural networks. This paper aims to establish the potential for removing the uncertainty surrounding the lack of user information when attracting them to a particular domain using a cascade of neural networks and opera-tional actions with data and knowledge between them. This study relies on neural networks and public data from social networks outside the internal targeted advertising systems of those social networks. The data were col-lected through interaction with the VKontakte social network API. We used binary classification to determine the competitors' domain and multi-class classification to determine the effectiveness of competitors' advertis-ing to encourage further customer acquisition.