Title :
Mining Multidimensional Data Using Clustering Techniques
Author :
Pagani, Marco ; Bordogna, Gloria ; Valle, Massimiliano
Author_Institution :
CNR-IDPA, Dalmine
Abstract :
We describe a novel data mining procedure to discover relevant associations in multidimensional data. The procedure applies hierarchical clustering to distinct pattern sets(views) of the same dataset and identifies the best partitions in the two dendrograms that exhibit the greatest correlation.Finally the most relevant associations between pattern sets characterizing the most correlated clusters in the identified partitions are discovered. An application of the procedure to identify association between compositional views and performance views of a dataset of materials is discussed.
Keywords :
data mining; pattern clustering; clustering techniques; distinct pattern sets; hierarchical clustering; multidimensional data mining; relevant associations discovery; Conferences; Data mining; Databases; Expert systems; Multidimensional systems;
Conference_Titel :
Database and Expert Systems Applications, 2007. DEXA '07. 18th International Workshop on
Conference_Location :
Regensburg
Print_ISBN :
978-0-7695-2932-5
DOI :
10.1109/DEXA.2007.112
