DOI: 10.17587/prin.7.195-201

Source Code Indivisible Parts Symptoms Extraction Based on Universal Intermediate and Further Equivalent Representations

A. A. Kovalevskiy, morskoyzmey@gmail.com, A. N. Pustygin, p2008an@ya.ru, Chelyabinsk State University, Chelyabinsk, 454001, Russian Federation

Corresponding author: Kovalevskiy Aleksey A., Postgraduate Student, Chelyabinsk State University, Chelyabinsk, 454001, Russian Federation, e-mail: morskoyzmey@gmail.com

In this article a source code symptoms extraction method is proposed, based on source code transformation to universal intermediate and then to an equivalent representation. The goal of the transformation is the subsequent classification of text parts in accordance with the developed types system. The analysis of open source projects is based on symptoms of the parts without binding to identifier names. For four open source projects there are statistical results to a groups symptoms clustering accordingly with a having functionally in pairs similar appointment submitted. The aim of this study was as an implementation of the source code selection given set of functional attributes parts as well as demonstration of the using possibilities of the proposed approach to the source text analysis. The test system types (glossary) was created according to heuristic reasons to described quantum algorithmic meaning.

Keywords: universal intermediate representation, equivalent representation, statical analysis, open source software, source code classification, source code indivisible symptoms

pp. 195–201

For citation:
Kovalevskiy A. A., Pustygin A. N. Source Code Indivisible Parts Symptoms Extraction Based on Universal Intermediate and Further Equivalent Representations, Programmnaya Ingeneria, 2016, vol. 7, no. 5, pp. 195—201.