A. Sleit, S. Al-Adaileh, N. Al-Omari, and H. Hurani
Data clustering, cluster labelling, cluster visualization, centroid-based labelling, cluster map
The increasing demand for data in many applications has led to the existence of large datasets causing the need for clustering techniques to become an important challenge. Several clustering algorithms exist for processing large dataset such as representative-point-based labelling (RPBL) and centroid-based labelling (CBL) which are based on the very rough distribution of cluster boundary. The cluster map (CM) algorithm improves the precession of clustering algorithms by allowing the user to define and redefine clusters. User interaction is the main problem with the CM algorithm because most systems do not allow user interaction. This paper proposes a cluster identifier (CI) algorithm which automates the user role in the CM algorithm. CI defines clusters, assigns a unique identifier for each cluster and labels the outliers with special identifiers to distinguish them from clusters. Experimental results demonstrate that automating user role in the CI algorithm produces comparable results to those generated by the CM algorithm.
Important Links:
Go Back
