Paper Title
A Study on Anomalous Cluster Detection in High Dimensional Discrete Data Sets over Social Media

Abstract
We propose an algorithm for anomalous detection of high dimensional discrete data usingan approach of clustering anomalies from the discrete data sets. Rather than the normal AD algorithm that detects the set of points which collectively exhibit abnormal patterns, providing a systematic way of detecting an anomaly with in an unanimity concern. The proposed algorithm emphasis efficient and powerful detection of anomalies. Unlike the existing techniques of finding each word separately the algorithm here uses a clustering method to detect each word that possess a maximum deviance from the normal pattern collectively in the batch of a given text document. Thereby resulting in more advantageous and effectual way of anomalous discovery over social networking sites preventing abnormal patterns. Index Terms�Anomaly Detection, Pattern Detection, Clustering method, Discrete data, Statistical analysis, Unethical pattern detection