An effective approach using dissimilarity measures to estimate software code clone

Software based clone detection is in hype as industries demand to such product has risen. Due to code replication means the copy and paste activities, such pattern is recurrent thereby developers can reduce effort and time of rewriting similar code fragment from scratch. In the industrial software system, code replication is found a serious trouble because it may affect on quality, consistency, maintainability and comprehensibility. Thus, efficient approach is needed to detect such replication in distributed environment. The trial here is variety of syntax, compiler dependent language, and various coding styles to solve a single problem. As per the related survey, researchers are finding difficult to evolve code copies, even on regressive benchmarking. The existing software tools have some restrictions to detect perfect code clone. Each software developer may think in different way for the implementation of the same problem. The methodology explained here is to specify an efficient way to detect code clone which is a hybrid model that covers maximum coding behavior and classes of clones. Along with similarity check, the paper describes the importance of dissimilarity detection. Detecting dissimilarity is due to operator or function overloading. Since this is essential feature of a good Object oriented Language. It also discusses key techniques that save time in retrieval and comparison of data, by extracting and arranging code that is mined from code document. The proposed system eliminates efforts of comparing the code line by line between two files, which was followed in traditional algorithm. It defines a reduction technique and code complexity based analysis which increases the probability of success. The concluding mark is that no single scheme defines procedure for all types of clone´s detection. In this paper, we introduce a multi-model learning technique to detect various types of code clone, which has been taken up as problem statement in this research wo- k.