Optimal Functionality and Domain Data Clustering Based on Latent Dirichlet Allocation

Date Added: Oct 2011
Format: PDF

This paper presents a new approach for clustering domain data and application functionality, based on the Latent Dirichlet Allocation. The methodology, developed here, performs an optimal clustering by identifying input values that lead to the best possible clustering output. The optimal solutions are identified through the use of the Silhouette technique. A validation of the work is performed based on the TPC-W benchmark. The new approach is flexible enough to be applied to any object-oriented application where identifying meaningful clusters of its domain data and functionality is desired.