Back RSS stream

Publications of Jérôme Darmont

Reference (inproceedings)

F. Bentayeb, O. Boussaïd, J. Darmont, "Multi-Link Lists as Data Cube Structures in the MOLAP Environment", 14th IRMA International Conference, Philadelphia, USA, May 2003, 35-37; IRMA, Hershey, USA.

BibTeX entry

     Author = {Fadila Bentayeb and Omar Boussaïd and Jérôme Darmont},
     Title = {Multi-Link Lists as Data Cube Structures in the MOLAP Environment},
     Booktitle = {14th IRMA International Conference, Philadelphia, USA},
     Month = {May},
     Year = {2003},
     Organization = {IRMA},
     Pages = {35-37},
     Abstract = {In the area of On Line Analytical Processing (OLAP), the concept of multidimensional databases is growing in popularity. OLAP indeed allows to model data in a multidimensional way (e.g., in a data cube) in order to look at the data from many different points of view. Several efficient algorithms for Relational OLAP (ROLAP) have been developed to compute the cube. Multidimensional OLAP (MOLAP) systems present a different challenge in computing the cube. The main difference resides in the data structures representing the data. ROLAP systems store data in relational tables while MOLAP systems store their data as sparse arrays. To increase the efficiency of arrays, many authors propose to chunk arrays to avoid space memory problem and compress data to avoid the problem of sparse data. Moreover, another problem in MOLAP systems is the lack of scalability. In this paper, we focused only on the data structure that stores multidimensional data in the MOLAP environment. Our objective is to provide a new data cube structure whose characteristics are interesting compared to fixed array data structures. Indeed, the data cube structure we propose is a set of multi-link lists (MLL) that is a dynamic data structure rather than fixed-sized arrays. With the MLL data structure, we (1) avoid the sparse data problem by represent-ing only real data, (2) facilitate navigation through the data for further OLAP operations, and (3) can easily extend the data cube when adding dimensions or hierarchies without re-build it. Furthermore, this MLL data structure can be itself considered as an index mechanism thanks to multiple links between data. Moreover, we show that the MLL data structure presents good performances in terms of space and query response time when compared to array structures.},
     Keywords = {Multidimensional data, Data cube, MOLAP, Sparse Arrays, Multi-Link Lists}

[ Export | Back ]