A hybrid approach (syntactic and textual) to clone detection

Abstract

Code clone detection has been so far tackled with several approaches (mainly textual, syntactic, and semantic), each performing best w.r.t. specific metrics and against consolidated but specific benchmarks. Hybrid approaches have been recently proposed as well. This paper proposes a novel hybrid (syntactic, textual) approach using the abstract syntax tree to identify clone candidates and textual methods to discard false positives. The novelty of the approach is in the combination of two well-grounded techniques, in a way that has not been explored yet. The paper shows the result of empirical evaluation methods and hints of a prototype implementation. Initial experimental results show that the approach is effective and can find clones that were not detected by state of the art approaches.