Knowledge Extraction by Dynamical Clustering of sea waves streaming data.

Data stream can be thought as a sequence of ordered data items, where the input arrives more or less continuously as time progress. There exist several applications producing data stream, e.g. telecommunication system, stock markets customer click streams etc. In this paper we consider the problem of extracting knowledge by a dinamical clustering algorithm of sea waves streaming data, that is to say evolving streaming of data coming from a multisensor system. For this purpose we develop an updating version of Dynamical Clustering Algorithm [5]. This problem is very interesting from a practical point of view. It is based on the computation of a prototypal wave through a free-knot smoothing spline, optimizing a non linear problem. Thanks to this approach, it is possible to investigate in which way the incoming data change according to the various steps of process registration, and to have a summary description of the entire data thought prototypals flowing curves using a small amount of memory and time.