CodeSum: Translate Program Language to Natural Language

During software maintenance, programmers spend a lot of time on code comprehension. Reading comments is an effective way for programmers to reduce the reading and navigating time when comprehending source code. Therefore, as a critical task in software engineering, code summarization aims to generate brief natural language descriptions for source code. In this paper, we propose a new code summarization model named CodeSum. CodeSum exploits the attention-based sequence-to-sequence (Seq2Seq) neural network with Structure-based Traversal (SBT) of Abstract Syntax Trees (AST). The AST sequences generated by SBT can better present the structure of ASTs and keep unambiguous. We conduct experiments on three large-scale corpora in different program languages, i.e., Java, C#, and SQL, in which Java corpus is our new proposed industry code extracted from Github. Experimental results show that our method CodeSum outperforms the state-of-the-art significantly. Introduction Source code summarization is the task of creating readable natural language summaries that describe the functionality of software. It is important in the field of source code comprehension. During the software maintenance, programmers spend a lot of time reading and understanding the source code snippets to comprehend them. Studies of program comprehension indicate that programmers often read a summary which is a comment describing the function of the code (e.g., JavaDoc1 descriptions for Java methods) or skim source code (e.g., read important keywords) to save time (Rodeghero et al. 2014; Sim, Clarke, and Holt 1998). For example, Figure 1 shows a Java method named toIndexName extracted from Github2. Through the summary and name of the method, developers can easily understand the method aiming to “convert the index of an attacker into a readable name in a battle”. However, these summaries are sometimes missing, incomplete or outdated. Therefore, automated source code summarization becomes an emerging technology in software engineering. Predicting these source code summarizations can be used in improving code search by natural language queries, code comprehension, and code