MPT‐embedding: An unsupervised representation learning of code for software defect prediction. Issue 4 (15th December 2020)

• Record Type: Journal Article
• Authors: Shi, Ke; Lu, Yang; Liu, Guangliang; Wei, Zhenchun; Chang, Jingfei
• Is Part Of: Journal of software. Volume 33:Issue 4(2021)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2020-12-15
• Subjects: deep learning -- defect prediction -- representation learning -- tree embedding; Software engineering -- Periodicals; Computer software -- Development -- Periodicals; Software maintenance -- Periodicals; 005.1
• DOI: 10.1002/smr.2330
• Languages: English
• ISSNs: 2047-7473

Abstract

Abstract: Software project defect prediction can help developers allocate debugging resources. Existing software defect prediction models are usually based on machine learning methods, especially deep learning. Deep learning‐based methods tend to build end‐to‐end models that directly use source code‐based abstract syntax trees (ASTs) as input. They do not pay enough attention to the front‐end data representation. In this paper, we propose a new framework to represent source code called multiperspective tree embedding (MPT‐embedding), which is an unsupervised representation learning method. MPT‐embedding parses the nodes of ASTs from multiple perspectives and encodes the structural information of a tree into a vector sequence. Experiments on both cross‐project defect prediction (CPDP) and within‐project defect prediction (WPDP) show that, on average, MPT‐embedding provides improvements over the state‐of‐the‐art method. Abstract : Source code‐based automatic representations are more objective and accurate than traditional handcrafted metrics. This article proposed a new framework to represent code called multiperspective tree embedding (MPT‐embedding), which is an unsupervised representation learning method. MPT‐embedding parses the nodes of ASTs from multiple perspectives and encodes the structural information of a tree into a vector sequence. Experiments on the tasks of defect prediction show the effectiveness of the model.