Semantically Enhanced Code Clone Refinement Algorithm Based on Analysis of Multiple Detection Reports

An algorithm based on semantic analysis of multiple detection tools' reports using WordNet is proposed oriented on the refinement of code clones. It parses different detection tools' reports looking for new clone specifications, and refines the location of existing ones using semantic information contained in source code. It is applied to a real and complex software system and is compared to three other well-known detection algorithms, discovering 4888 clone pairs more than the average detected by other tools; also making the code clones 3 lines longer (for a subset of the same system the results are proportional to the size reduction). The objective is to provide higher quantity of code clones, and more appropriated localization to be used in refactoring processes.