ABSTRACT

Clustering is a data mining technique of grouping similar type of data or queries together which helps in identifying similar subject areas. The major problem is to identify heterogeneous subject areas where frequent queries are asked. There are number of agglomerative clustering algorithms which are used to cluster the data. The problem with these algorithms is that they make use of distance measures to calculate similarity. So the best suited algorithm for clustering the categorical data is Robust Clustering Using Links (ROCK) [1] algorithm because it uses Jaccard coefficient instead of using the distance measures to find the similarity between the data or documents to classify the clusters. The mechanism for classifying the clusters based on the similarity measure shall be used over a given set of data. This method will make clusters of the data corresponding to different subject areas so that a prior knowledge about similarity can be maintained which in turn will help to discover accurate and consistent clusters and will reduce the query response time. The main objective of our work is to implement ROCK [1] and to decrease the query response time by searching the documents in the resulted clusters instead of searching the whole database. This technique actually reduces the searching time of documents from the database.