Context Free Data Cleaning and its Application in Mechanism for Suggestive Data Cleaning

Organizations are being flooded with massive transactional data. This data is of no use, if not analysed properly, to reach to any strategic decision and ultimately to achieve competitive advantage. The efficient data analysis is one of the success strategies. The analysis is highly dependent on the quality of the data. The clean data will lead to efficient data analysis. In this paper, authors suggest application of similarity metrics in context free data cleaning and a mechanism to suggest correct data based on learning from patterns derived in the prior phase. The sequence similarity metrics like Needlemen-Wunch, Jaro-Winkler, Chapman Ordered Name Compound Similarity and Smit-Watermen are used to find distance of two values. Experimental results show that how the approach not only effectively cleaning the data but suggesting suitable values in order to reduce the data entry errors.