论文信息 - Classification of source code solutions based on the solved programming tasks

Classification of source code solutions based on the solved programming tasks

The programming is based on the source code. It is often difficult to determine from the source code what task was solved, due to the diversity of programming. For this reason, developers often do not recognize the functionality of existing codes and waste time by re-implementing the problem. This paper presents a machine-based model that can effectively identify and assign the source code to the solved task. Machine learning implements by a convolution network, which requires the presence of consistent (properly functioning, outputs the expected output for each input) source codes to the input. To convert the text data will be used for text processing using CBOW model, which model will be able to classify the source codes written in the same language.

Sándor Szénási | Ádám Pintér

[1] Minh-Le Nguyen,et al. Sentence Modeling with Deep Neural Architecture using Lexicon and Character Attention Mechanism for Sentiment Classification , 2017, IJCNLP.

[2] Yoon Kim,et al. Convolutional Neural Networks for Sentence Classification , 2014, EMNLP.

[3] Zhi Jin,et al. Building Program Vector Representations for Deep Learning , 2014, KSEM.

[4] Jeffrey Dean,et al. Efficient Estimation of Word Representations in Vector Space , 2013, ICLR.

[5] Mohammad Ebrahim Khademi,et al. Conceptual Text Summarizer: A new model in continuous vector space , 2017, ArXiv.

[6] Z. Duric,et al. A Source Code Similarity System for Plagiarism Detection , 2013, Comput. J..

[7] Zhuowen Tu,et al. Generalizing Pooling Functions in Convolutional Neural Networks: Mixed, Gated, and Tree , 2015, AISTATS.

[8] Yijun Yu,et al. Cross-Language Learning for Program Classification using Bilateral Tree-Based Convolutional Neural Networks , 2017, AAAI Workshops.

[9] Shlok Gilda,et al. Source code classification using Neural Networks , 2017, 2017 14th International Joint Conference on Computer Science and Software Engineering (JCSSE).