A Sliding Window Algorithm for Relational Frequent Patterns Mining from Data Streams

Some challenges in frequent pattern mining from data streams are the drift of data distribution and the computational efficiency. In this work an additional challenge is considered: data streams describe complex objects modeled by multiple database relations. A multi-relational data mining algorithm is proposed to efficiently discover approximate relational frequent patterns over a sliding time window of a complex data stream. The effectiveness of the method is proved on application to the Internet packet stream.

• Type: Conference
• Year: 2009
• Where: DIS
• Authors: Fabio Fumarola, Anna Ciampi, Annalisa Appice, Donato Malerba
• Keywords: Data Stream, DIS 2009, Frequent Pattern Mining, Frequent Patterns, Theoretical Computer Science