International Journal of Application or Innovation in Engineering & Management (IJAIEM)
In this paper the authors provide a way to deal with the presentation faults of the traditional linear file format for data sets from databases, especially in database clustering. A better approach called extensive data set that allow attributes of an object to have multiple values and it is shown how enhanced extensive data set can represent structural information in databases for clustering. Analyzing the problems of linear file format this extensive data set proves to be better. A unified similarity measure framework is proposed for single valued and multi-valued attributes.