Абстрактный

An Approach to Improve Computer Forensic Analysis via Document Clustering Algorithms

J. Shankar Babu, K.Sumathi

In computer Forensic analysis thousands of files are usually examined. The computer examiners feel much difficult to analyze data in those files, since it consists of only unstructured text or information. The majority of the tools available on the market have the ability to permit investigators to analyze the information or data that was gathered from a computer Systems. In this context, forensic analysis plays a major role by examining suspected documents seized in police investigations. We proposed an approach clustering algorithms to estimate the number of clusters formed while analyzing the document. New and useful knowledge is discovered while clustering the documents by our algorithm. K-means, K-medoids, Single Link, Complete Link, Average Link, and Cluster-based Similarity Partitioning Algorithm (CSPA) aredifferent efficient algorithms used for clustering documents. To find the number of clusters formed, we use two relative validity indexes in our approach. As a final point, we present several practical results of forensic computing.