The process of automatic extraction, recognition, description and classification of patterns from huge amount of data plays an important role in modern volcano monitoring techniques. In particular, the ability of certain systems to recognize different volcano status can help the researchers to better understand the complex dynamics underlying the geophysical system. The geophysical data are automatically measured and recorded by geophysical instruments. Their interpretation is very important for the investigation of earth s behavior. The fundamental task of volcano monitoring is to follow volcanic activity and promptly recognize any changes. To achieve such goals, different geophysical techniques (i.e. seismology, ground deformation, remote sensing, magnetic and electromagnetic studies, gravimetric) are used to obtain precise measurements of the variations induced by an evolving magmatic system. To proper exploit the wealth of such heterogeneous data, algorithms and techniques of data mining are fundamental tools. This thesis can be considered a detailed report about the application of the data mining discipline in the geophysical area. After introducing the basic concepts and the most important techniques constituting the state-of-art in the data mining field, we will apply several methods able to reach important results about the extraction of unknown recurrent patterns in seismic and infrasonic signals, and we will show the implementation of systems representing efficient tools for the monitoring purpose.