Le Monde des Utilisateurs de L'Analyse de Données

Numéro 36


Summarizing a 3 Way Relational Data Stream. Baptiste CSERNEL, Fabrice CLÉROT and Georges HÉBRAIL
La revue MODULAD, numéro 36, Juillet 2007

In this paper, we present a novel method to summarize a group of three data streams sharing a relational link between each other. That is, using the relational data base model as a reference, two entity streams and one stream of relationships. Each entity stream contains a stream of entities, each one of them referenced by a key, much in the same way as a primary key references all objects in a regular database table. The stream of relationships contains key couples identifying links between objects in the two entity streams as well as attributes characterising this particular link.
The algorithm presented here produces not only a summary of both entity streams considered independently of each other, but also gives a summary of the relations existing between the two entity streams as well as information on the join obtained joining the two entity streams through the relationship stream.

Keywords: Relational DataMining, Data Stream Mining, Clustering, Data Stream
Summaries, Data Stream Joins.

Download paper : Summarizing a 3 Way Relational Data Stream

Download slides : Summarizing a 3 Way Relational Data Stream