Stream-based live entity resolution approach with adaptive duplicate count strategy

Recently, researchers have been more concerned about large-scale news and tweet data generated by the social media. Some cloud service providers utilise the data to find public sentiments for the tenants. The challenge is how to clean the big data in the cloud before making further analysis. To address this issue, we propose a new live entity resolution approach at a time to find duplicates from the news and tweet data. We investigate possible solutions to address live entity resolution in the cloud, to make sliding window size adaptive using multistep distance and window size dependent duplicate count strategy with alterable window step, and find duplicates by overlapping boundary objects in adjacent blocks. Finally, our experimental evaluation based on the news data on large datasets shows the high effectiveness and efficiency of the proposed approaches.