The project goal is to find a good algorithm to learn the class of Spam SMS's. One of the research questions was what are the most suitable attributes in order to learn the class. We have tried two different learning algorithm: Adaboost and K-Nearest Neighbor. In this paper we are presenting our work: the dataset used, the benchmarks done and our conclusion. The paper includes also a discussion about how the ratio between the different classes in the dataset can affect the results of the benchmarks.