Clustering Users’ Requirements Schemas
Nouha Arfaoui, Jalel Akaichi · 2014
Data Mining proposes different techniques to deal with data. In our work, we suggest the use of clustering technique since we want grouping the schemas into clusters according to their similarity. This technique is applied to variety type of variables. We focus on categorical data. Many algorithms are proposed, but no one of them takes into consideration the semantic aspect. For this reason, and in order to ensure a good clustering of the schemas of the users’ requirements, we extend the k-mode algorithm by modifying its dissimilarity measure. The schemas within each cluster will be merged to construct the schemas of the data mart.