Software Clone Detection Using Clustering Approach

Code clones are highly similar or identical code segments. Identification of clones helps improve software quality through managed evolution, refactoring, complexity reduction, etc. In this study, we investigate Type 1 and Type 2 function clones using a data mining technique. First, we create a dataset by collecting metrics for all functions in a software system. Second, we apply DBSCAN clustering algorithm on the dataset so that each cluster can be analysed to detect Type 1 and Typei¾ź2 function clones. We evaluate our approach by analyzing an open source software Bitmessage. We calculate the precision value to show the effectiveness of our approach in detecting function clones. We show that our approach for functional clone detection is effective with high precision value and number of function clones detected.