Levenshtein edit distance-based type III clone detection using metric trees

This paper presents an original technique for clone detection with metric trees using Levenshtein distance as the metric defined between two code fragments. This approach achieves a faster empirical performance. The resulting clones may be found with varying thresholds allowing type 3 clone detection. Experimental results of metric trees performance as well as clone detection statistics on an open source system are presented and give promising perspectives.