Mining patterns from large star schemas based on streaming algorithms

Abstract

A growing challenge in data mining is the ability to deal with complex, voluminous and dynamic data. In many real world applications, complex data is not only organized in multiple database tables, but it is also continuously and endlessly arriving in the form of streams. Although there are some algorithms for mining multiple relations, as well as a lot more algorithms to mine data streams, very few combine the multi-relational case with the data streams case. In this paper we describe a new algorithm, Star FP-Stream, for finding frequent patterns in multi-relational data streams following a star schema. Experiments in the emphAdventureWorks data warehouse show that Star FP-Stream is accurate and performs better than the equivalent algorithm, FP-Streaming, for mining patterns in a single data stream. © 2012 Springer-Verlag Berlin Heidelberg.