Improving similarity join algorithms using vertical clustering techniques

String is a primary data format in majority of applications. With the rapid growth of diverse data driven applications in the current information era, retrieving string data from heterogeneous structured sources becomes more and more significant and challenging. The main concern is duplicate records are created when data is integrated from heterogeneous sources. Those duplicate records represent the same real-world entity because of inconsistent values and naming conventions, incorrect or missing data values, or incomplete information. Existing approaches make the assumption that group of related attributes will participate in the similarity join operation. However, in this paper we propose a pre-processing technique to improve existing similarity join techniques. Assuming relational data sources, our approach is to identify groups of related attributes that when similarity join is applied, we reduce false positives and false negatives, and increase precisions and F-measure.