On the Use of Automated Text Summarization Techniques for Summarizing Source Code

By Sonia Haiduc, Jairo Aponte, Laura Moreno, Andrian Marcus. Working Conference on Reverse Engineering, 2010

429 citations recorded by the source corpus.

Subfields: Computer Networks and Communications, Information Systems

View Paper: https://www.semanticscholar.org/paper/909f24eb765764fdc5622e1f8c121205ce93db1b

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Networks and Communications

In Information Systems

Citations in the corpus

Deep Code Comment Generation (2018, methodology): “Automatic code summarization approaches vary from manuallycrafted template [24, 35, 36], IR [14, 15, 43] to learning-based approaches [4, 19, 28].”
Improved Code Summarization via a Graph Neural Network (2020, methodology): “ manual e‡ort of writing summaries themselves [12, 26, 45, 48], and automation has long been cited as a desirable alternative [17]. „e term “source code summarization” was coined around ten years ago [19] and since that time the •eld has proliferated. At •rst, the dominant strategy was based on sent”
Using machine translation for converting Python 2 to Python 3 code (2015, methodology): “Haiduc et al. (2010) have worked on summarizing the source code by using Latent Semantic In-dexing and Vector Space Model to generate summaries for a java code.”
CodeScope: An Execution-based Multilingual Multitask Multidimensional Benchmark for Evaluating LLMs on Code Understanding and Generation (2023, positive): “5 also performs well, achieving a significant improvement of 76.67% in memory optimization for C language.”
Learning to Generate Pseudo-Code from Source Code Using Statistical Machine Translation (T) (2015, neutral): “1 is one example of the type of parallel corpus targeted in this study, in which have one-by-one correspondences between each line in the source code and pseudo-code.”
Retrieval-based Neural Source Code Summarization (2020, methodology): “As described above, IR-based methods can effectively leverage the existing terms of the original code or summaries of similar code snippets [13, 18, 19, 46, 60, 61], while NMT-based methods generate summaries word-by-word from the whole corpus by directly maximizing the likelihood of the next word…”
TranS^3: A Transformer-based Framework for Unifying Code Summarization and Code Search (2020, neutral): “[64] combined IR techniques, i.”
Autofolding for Source Code Summarization (2014, methodology): “words [14], [15], [16], [17]; or (d) (potentially discontiguous) lines of code that match a user’s query [18].”
Studying the Usage of Text-To-Text Transfer Transformer to Support Code-Related Tasks (2021, methodology): “The proposed summarization techniques fall into two categories: extractive [43]– [46] and abstractive [9], [11], [47]–[49].”
CoCoGUM: Contextual Code Summarization with Multi-Relational GNN on UMLs (2020, methodology): “Haiduc et al. [5, 55] use the TF-IDF metric to rank the terms in a method body based on the importance those terms are to that method.”
Suggesting accurate method and class names (2015, positive): “Haiduc et al. showed that NL text summarization does not work well for code [23] and such techniques must be adapted to be effective.”
Learning to Spot and Refactor Inconsistent Method Names (2019, methodology): “[83] used natural language summarization techniques and the lexical and structural context in code to improve code comprehension [84].”
A Neural Model for Generating Natural Language Summaries of Program Subroutines (2019, background): “2) Heuristic/Template-based: Haiduc et al. [11], [36] is often cited as the ﬁrst attempt to create text summaries of code, and indeed is the ﬁrst to introduce the term “source code summarization.””
Automatic generation of natural language summaries for Java classes (2013, methodology): “While successfully applied for Java methods [4], generating comments for more complex code artifacts, e.g., classes, is significantly more difficult [5] [6].”
RepoAgent: An LLM-Powered Open-Source Framework for Repository-level Code Documentation Generation (2024, neutral): “Early methods were rule-based or template-driven (Haiduc et al., 2010; Sridhara et al., 2010; Moreno et al., 2013; Rodeghero et al., 2014).”
Automatically generating commit messages from diffs using neural machine translation (2017, methodology): “Extractive summarization extracts relevant parts of source code and uses the relevant parts as a summary [16].”