In this article we present an approach in predicting a user's behavior using big data, correlations, Ramsey theory and generalized suffix trees. Recommendation systems are widely used, especially in e-commerce. The purpose of the recommendation system is to predict what items a user would prefer. This predictions are based on some information like the group in which the user could fit or the past behavior of the user (for example last checked items). The more data we have, the best predictions we obtain. In last few years, it was collected so much information, which could not be stored in classical database management systems, so we already are talking about big data. Firstly, we present the main concepts regarding recommendation systems and big data, then we present some mathematical concepts and in the last part, we analyze the mathematical results and try to apply them in big data and recommendation systems, using generalized suffix trees.