The paper deals with detection of structures in source codes employing statistical classification. To enhance source code perception by development tools like code editors, modeling tools and source code repositories, various methods of patterns classification are proposed and tested. To be able to apply classification algorithms, well-defined feature space is required. Thus, such a feature space is presented and tested. Sub-models search is carried out by a genetic algorithm to select the optimal feature space subset without deterioration of a classification system. The results show that with standard classification algorithms like k-NN or Perceptron, accuracy of 0.8 can be achieved.
CITATION STYLE
Mojzes, M., Rost, M., Smolka, J., & Virius, M. (2017). Application of statistical classifiers on Java source code. In Advances in Intelligent Systems and Computing (Vol. 511 AISC, pp. 208–218). Springer Verlag. https://doi.org/10.1007/978-3-319-46535-7_16
Mendeley helps you to discover research relevant for your work.