Shangwen Wang

Haoye Tian, Xunzhu Tang, Andrew Habib et al.

In this work, we propose a novel perspective to the problem of patch correctness assessment: a correct patch implements changes that "answer" to a problem posed by buggy behaviour. Concretely, we turn the patch correctness assessment into a Question Answering problem. To tackle this problem, our intuition is that natural language processing can provide the necessary representations and models for assessing the semantic correlation between a bug (question) and a patch (answer). Specifically, we consider as inputs the bug reports as well as the natural language description of the generated patches. Our approach, Quatrain, first considers state of the art commit message generation models to produce the relevant inputs associated to each generated patch. Then we leverage a neural network architecture to learn the semantic correlation between bug reports and commit messages. Experiments on a large dataset of 9135 patches generated for three bug datasets (Defects4j, Bugs.jar and Bears) show that Quatrain can achieve an AUC of 0.886 on predicting patch correctness, and recalling 93% correct patches while filtering out 62% incorrect patches. Our experimental results further demonstrate the influence of inputs quality on prediction performance. We further perform experiments to highlight that the model indeed learns the relationship between bug reports and code change descriptions for the prediction. Finally, we compare against prior work and discuss the benefits of our approach.

Xiaopeng Li, Shasha Li, Shangwen Wang et al.

Knowledge editing has emerged as an efficient technique for updating the knowledge of large language models (LLMs), attracting increasing attention in recent years. However, there is a lack of effective measures to prevent the malicious misuse of this technique, which could lead to harmful edits in LLMs. These malicious modifications could cause LLMs to generate toxic content, misleading users into inappropriate actions. In front of this risk, we introduce a new task, $\textbf{K}$nowledge $\textbf{E}$diting $\textbf{T}$ype $\textbf{I}$dentification (KETI), aimed at identifying different types of edits in LLMs, thereby providing timely alerts to users when encountering illicit edits. As part of this task, we propose KETIBench, which includes five types of harmful edits covering the most popular toxic types, as well as one benign factual edit. We develop five classical classification models and three BERT-based models as baseline identifiers for both open-source and closed-source LLMs. Our experimental results, across 92 trials involving four models and three knowledge editing methods, demonstrate that all eight baseline identifiers achieve decent identification performance, highlighting the feasibility of identifying malicious edits in LLMs. Additional analyses reveal that the performance of the identifiers is independent of the reliability of the knowledge editing methods and exhibits cross-domain generalization, enabling the identification of edits from unknown sources. All data and code are available in https://github.com/xpq-tech/KETI.

Xiaohu Du, Ming Wen, Zichao Wei et al.

Transformer-based pre-trained models of code (PTMC) have been widely utilized and have achieved state-of-the-art performance in many mission-critical applications. However, they can be vulnerable to adversarial attacks through identifier substitution or coding style transformation, which can significantly degrade accuracy and may further incur security concerns. Although several approaches have been proposed to generate adversarial examples for PTMC, the effectiveness and efficiency of such approaches, especially on different code intelligence tasks, has not been well understood. To bridge this gap, this study systematically analyzes five state-of-the-art adversarial attack approaches from three perspectives: effectiveness, efficiency, and the quality of generated examples. The results show that none of the five approaches balances all these perspectives. Particularly, approaches with a high attack success rate tend to be time-consuming; the adversarial code they generate often lack naturalness, and vice versa. To address this limitation, we explore the impact of perturbing identifiers under different contexts and find that identifier substitution within for and if statements is the most effective. Based on these findings, we propose a new approach that prioritizes different types of statements for various tasks and further utilizes beam search to generate adversarial examples. Evaluation results show that it outperforms the state-of-the-art ALERT in terms of both effectiveness and efficiency while preserving the naturalness of the generated adversarial examples.

Zhensu Sun, Xiaoning Du, Fu Song et al.

Currently, large pre-trained language models are widely applied in neural code completion systems. Though large code models significantly outperform their smaller counterparts, around 70\% of displayed code completions from Github Copilot are not accepted by developers. Being reviewed but not accepted, their help to developer productivity is considerably limited and may conversely aggravate the workload of developers, as the code completions are automatically and actively generated in state-of-the-art code completion systems as developers type out once the service is enabled. Even worse, considering the high cost of the large code models, it is a huge waste of computing resources and energy, which severely goes against the sustainable development principle of AI technologies. However, such waste has never been realized, not to mention effectively addressed, in the research community for neural code completion. Hence, preventing such unhelpful code completions from happening in a cost-friendly way is of urgent need. To fill this significant gap, we first investigate the prompts of unhelpful code completions, called "low-return prompts". We empirically identify four observable patterns in low-return prompts, each lacking necessary information, making it difficult to address through enhancements to the model's accuracy alone. This demonstrates the feasibility of identifying such low-return prompts based on the prompts themselves. Motivated by this finding, we propose an early-rejection mechanism to turn down low-return prompts by foretelling the code completion qualities. The prompts that are estimated to receive unhelpful code completions will not be sent to the model. Furthermore, we investigated five types of estimators to demonstrate the feasibility of the mechanism. The experimental results show that the estimator can reject 20% of code completion requests with a 97.4% Precision.

Xiaopeng Li, Shasha Li, Xi Wang et al.

Large Language Models (LLMs) power numerous AI applications, yet updating their knowledge remains costly. Model editing provides a lightweight alternative through targeted parameter modifications, with meta-learning-based model editing (MLME) demonstrating strong effectiveness and efficiency. However, we find that MLME struggles in low-data regimes and incurs high training costs due to the use of KL divergence. To address these issues, we propose $\textbf{E}$fficient $\textbf{M}$ulti-$\textbf{S}$tep $\textbf{Edit (EMSEdit)}$, which leverages multi-step backpropagation (MSBP) to effectively capture gradient-activation mapping patterns within editing samples, performs multi-step edits per sample to enhance editing performance under limited data, and introduces norm-based regularization to preserve unedited knowledge while improving training efficiency. Experiments on two datasets and three LLMs show that EMSEdit consistently outperforms state-of-the-art methods in both sequential and batch editing. Moreover, MSBP can be seamlessly integrated into existing approaches to yield additional performance gains. Further experiments on a multi-hop reasoning editing task demonstrate EMSEdit's robustness in handling complex edits, while ablation studies validate the contribution of each design component. Our code is available at https://github.com/xpq-tech/emsedit.

Kui Liu, Shangwen Wang, Anil Koyuncu et al.

Test-based automated program repair has been a prolific field of research in software engineering in the last decade. Many approaches have indeed been proposed, which leverage test suites as a weak, but affordable, approximation to program specifications. Although the literature regularly sets new records on the number of benchmark bugs that can be fixed, several studies increasingly raise concerns about the limitations and biases of state-of-the-art approaches. For example, the correctness of generated patches has been questioned in a number of studies, while other researchers pointed out that evaluation schemes may be misleading with respect to the processing of fault localization results. Nevertheless, there is little work addressing the efficiency of patch generation, with regard to the practicality of program repair. In this paper, we fill this gap in the literature, by providing an extensive review on the efficiency of test suite based program repair. Our objective is to assess the number of generated patch candidates, since this information is correlated to (1) the strategy to traverse the search space efficiently in order to select sensical repair attempts, (2) the strategy to minimize the test effort for identifying a plausible patch, (3) as well as the strategy to prioritize the generation of a correct patch. To that end, we perform a large-scale empirical study on the efficiency, in terms of quantity of generated patch candidates of the 16 open-source repair tools for Java programs. The experiments are carefully conducted under the same fault localization configurations to limit biases.

Shangwen Wang, Xiaoguang Mao, Yue Yu

Organ transplantation, which is the utilization of codes directly related to some specific functionalities to complete ones own program, provides more convenience for developers than traditional component reuse. However, recent techniques are challenged with the lack of organs for transplantation. Hence, we conduct an empirical study on extracting organs from GitHub repository to explore transplantation based on large-scale dataset. We analyze statistics from 12 representative GitHub projects and get the conclusion that 1) there are abundant practical organs existing in commits with add as a key word in the comments; 2) organs in this repository mainly possess four kinds of contents; 3) approximately 70% of the organs are easy-to-transplant. Implementing our transplantation strategy for different kinds of organs, we manually extract 30 organs in three different programming languages, namely Java, Python, and C, and make unit tests for them utilizing four testing tools (two for Java, one for Python, and one for C). At last, we transplant three Java organs into a specific platform for a performance check to verify whether they can work well in the new system. All the 30 organs extracted by our strategy possess good performances in unit test with the highest passing rate reaching 97% and the lowest one still passing 80% and the three Java organs work well in the new system, providing three new functionalities for the host. All the results indicate the feasibility of organ transplantation based on open-source repository, bringing new idea for code reuse.

Tanghaoran Zhang, Yue Yu, Xinjun Mao et al.

Code snippet adaptation is a fundamental activity in the software development process. Unlike code generation, code snippet adaptation is not a "free creation", which requires developers to tailor a given code snippet in order to fit specific requirements and the code context. Recently, large language models (LLMs) have confirmed their effectiveness in the code generation task with promising results. However, their performance on adaptation, a reuse-oriented and context-dependent code change prediction task, is still unclear. To bridge this gap, we conduct an empirical study to investigate the performance and issues of LLMs on the adaptation task. We first evaluate the adaptation performances of three popular LLMs and compare them to the code generation task. Our result indicates that their adaptation ability is weaker than generation, with a nearly 15% decrease on pass@1 and more context-related errors. By manually inspecting 200 cases, we further investigate the causes of LLMs' sub-optimal performance, which can be classified into three categories, i.e., Unclear Requirement, Requirement Misalignment and Context Misapplication. Based on the above empirical research, we propose an interactive prompting approach to eliciting LLMs' adaptation ability. Experimental result reveals that our approach greatly improve LLMs' adaptation performance. The best-performing Human-LLM interaction successfully solves 159 out of the 202 identified defects and improves the pass@1 and pass@5 by over 40% compared to the initial instruction-based prompt. Considering human efforts, we suggest multi-agent interaction as a trade-off, which can achieve comparable performance with excellent generalization ability. We deem that our approach could provide methodological assistance for autonomous code snippet reuse and adaptation with LLMs.

Kang Yang, Xinjun Mao, Shangwen Wang et al.

Pre-trained code models rely heavily on high-quality pre-training data, particularly human-written reference comments that bridge code and natural language. However, these comments often become outdated as software evolves, degrading model performance. Large language models (LLMs) excel at generating high-quality code comments. We investigate whether replacing human-written comments with LLM-generated ones improves pre-training datasets. Since standard metrics cannot assess reference comment quality, we propose two novel reference-free evaluation tasks: code-comment inconsistency detection and semantic code search. Results show that LLM-generated comments are more semantically consistent with code than human-written ones, as confirmed by manual evaluation. Leveraging this finding, we rebuild the CodeSearchNet dataset with LLM-generated comments and re-pre-train CodeT5. Evaluations demonstrate that models trained on LLM-enhanced data outperform those using original human comments in code summarization, generation, and translation tasks. This work validates rebuilding pre-training datasets with LLMs to advance code intelligence, challenging the traditional reliance on human reference comments.

Zhensu Sun, Xiaoning Du, Fu Song et al.

Leveraging recent advancements in large language models, modern neural code completion models have demonstrated the capability to generate highly accurate code suggestions. However, their massive size poses challenges in terms of computational costs and environmental impact, hindering their widespread adoption in practical scenarios. Dynamic inference emerges as a promising solution, as it allocates minimal computation during inference while maintaining the model's performance. In this research, we explore dynamic inference within the context of code completion. Initially, we conducted an empirical investigation on GPT-2, focusing on the inference capabilities of intermediate layers for code completion. We found that 54.4% of tokens can be accurately generated using just the first layer, signifying significant computational savings potential. Moreover, despite using all layers, the model still fails to predict 14.5% of tokens correctly, and the subsequent completions continued from them are rarely considered helpful, with only a 4.2% Acceptance Rate. These findings motivate our exploration of dynamic inference in code completion and inspire us to enhance it with a decision-making mechanism that stops the generation of incorrect code. We thus propose a novel dynamic inference method specifically tailored for code completion models. This method aims not only to produce correct predictions with largely reduced computation but also to prevent incorrect predictions proactively. Our extensive evaluation shows that it can averagely skip 1.7 layers out of 16 layers in the models, leading to an 11.2% speedup with only a marginal 1.1% reduction in ROUGE-L.

Ying Xiao, Shangwen Wang, Sicen Liu et al.

Software built on top of machine learning algorithms is becoming increasingly prevalent in a variety of fields, including college admissions, healthcare, insurance, and justice. The effectiveness and efficiency of these systems heavily depend on the quality of the training datasets. Biased datasets can lead to unfair and potentially harmful outcomes, particularly in such critical decision-making systems where the allocation of resources may be affected. This can exacerbate discrimination against certain groups and cause significant social disruption. To mitigate such unfairness, a series of bias-mitigating methods are proposed. Generally, these studies improve the fairness of the trained models to a certain degree but with the expense of sacrificing the model performance. In this paper, we propose FITNESS, a bias mitigation approach via de-correlating the causal effects between sensitive features (e.g., the sex) and the label. Our key idea is that by de-correlating such effects from a causality perspective, the model would avoid making predictions based on sensitive features and thus fairness could be improved. Furthermore, FITNESS leverages multi-objective optimization to achieve a better performance-fairness trade-off. To evaluate the effectiveness, we compare FITNESS with 7 state-of-the-art methods in 8 benchmark tasks by multiple metrics. Results show that FITNESS can outperform the state-of-the-art methods on bias mitigation while preserve the model's performance: it improved the model's fairness under all the scenarios while decreased the model's performance under only 26.67% of the scenarios. Additionally, FITNESS surpasses the Fairea Baseline in 96.72% cases, outperforming all methods we compared.

Shangwen Wang, Kui Liu, Bo Lin et al.

Software Fault Localization refers to the activity of finding code elements (e.g., statements) that are related to a software failure. The state-of-the-art fault localization techniques, however, produce coarse-grained results that can deter manual debugging or mislead automated repair tools. In this work, we focus specifically on the fine-grained identification of code elements (i.e., tokens) that must be changed to fix a buggy program: we refer to it as fix localization. This paper introduces a neural network architecture (named Beep) that builds on AST paths to predict the buggy code element as well as the change action that must be applied to repair a program. Leveraging massive data of bugs and patches within the CoCoNut dataset, we trained a model that was (1) effective in localizing the buggy tokens with the Mean First Rank significantly higher than a statistics based baseline and a machine learning-based baseline, and (2) effective in predicting the repair operators (with the associated buggy code elements) with a Recall@1= 30-45% and the Mean First Rank=7-12 (evaluated by CoCoNut, ManySStuBs4J, and Defects4J datasets). To showcase how fine-grained fix localization can help program repair, we employ it in two repair pipelines where we use either a code completion engine to predict the correct token or a set of heuristics to search for the suitable donor code. A key strength of accurate fix localization for program repair is that it reduces the chance of patch overfitting, a challenge in generate-and-validate automated program repair: both two repair pipelines achieve a correctness ratio of 100%, i.e., all generated patches are found to be correct. Moreover, accurate fix localization helps enhance the efficiency of program repair.

Steffen Herbold, Alexander Trautsch, Benjamin Ledel et al.

Context: Tangled commits are changes to software that address multiple concerns at once. For researchers interested in bugs, tangled commits mean that they actually study not only bugs, but also other concerns irrelevant for the study of bugs. Objective: We want to improve our understanding of the prevalence of tangling and the types of changes that are tangled within bug fixing commits. Methods: We use a crowd sourcing approach for manual labeling to validate which changes contribute to bug fixes for each line in bug fixing commits. Each line is labeled by four participants. If at least three participants agree on the same label, we have consensus. Results: We estimate that between 17% and 32% of all changes in bug fixing commits modify the source code to fix the underlying problem. However, when we only consider changes to the production code files this ratio increases to 66% to 87%. We find that about 11% of lines are hard to label leading to active disagreements between participants. Due to confirmed tangling and the uncertainty in our data, we estimate that 3% to 47% of data is noisy without manual untangling, depending on the use case. Conclusion: Tangled commits have a high prevalence in bug fixes and can lead to a large amount of noise in the data. Prior research indicates that this noise may alter results. As researchers, we should be skeptics and assume that unvalidated data is likely very noisy, until proven otherwise.

Shangwen Wang, Ming Wen, Liqian Chen et al.

Background: Over the years, Automated Program Repair (APR) has attracted much attention from both academia and industry since it can reduce the costs in fixing bugs. However, how to assess the patch correctness remains to be an open challenge. Two widely adopted ways to approach this challenge, including manually checking and validating using automated generated tests, are biased (i.e., suffering from subjectivity and low precision respectively). Aim: To address this concern, we propose to conduct an empirical study towards understanding the correct patches that are generated by existing state-of-the-art APR techniques, aiming at providing guidelines for future assessment of patches. Method: To this end, we first present a Literature Review (LR) on the reported correct patches generated by recent techniques on the Defects4J benchmark and collect 177 correct patches after a process of sanity check. We investigate how these machine-generated correct patches achieve semantic equivalence, but syntactic difference compared with developer-provided ones, how these patches distribute in different projects and APR techniques, and how the characteristics of a bug affect the patches generated for it. Results: Our main findings include 1) we do not need to fix bugs exactly like how developers do since we observe that 25.4% (45/177) of the correct patches generated by APR techniques are syntactically different from developer-provided ones; 2) the distribution of machine-generated correct patches diverges for the aspects of Defects4J projects and APR techniques; and 3) APR techniques tend to generate patches that are different from those by developers for bugs with large patch sizes. Conclusion: Our study not only verifies the conclusions from previous studies but also highlights implications for future study towards assessing patch correctness.

Shangwen Wang, Ming Wen, Xiaoguang Mao et al.

Automated program repair (APR) has attracted widespread attention in recent years with substantial techniques being proposed. Meanwhile, a number of benchmarks have been established for evaluating the performances of APR techniques, among which Defects4J is one of the most wildly used benchmark. However, bugs in Mockito, a project augmented in a later-version of Defects4J, do not receive much attention by recent researches. In this paper, we aim at investigating the necessity of considering Mockito bugs when evaluating APR techniques. Our findings show that: 1) Mockito bugs are not more complex for repairing compared with bugs from other projects; 2) the bugs repaired by the state-of-the-art tools share the same repair patterns compared with those patterns required to repair Mockito bugs; however, 3) the state-of-the-art tools perform poorly on Mockito bugs (Nopol can only correctly fix one bug while SimFix and CapGen cannot fix any bug in Mockito even if all the buggy locations have been exposed). We conclude from these results that existing APR techniques may be overfitting to their evaluated subjects and we should consider Mockito, or even more bugs from other projects, when evaluating newly proposed APR techniques. We further find out a unique repair action required to repair Mockito bugs named external package addition. Importing the external packages from the test code associated with the source code is feasible for enlarging the search space and this action can be augmented with existing repair actions to advance existing techniques.

Shangwen Wang, Xiaoguang Mao, Nan Niu et al.

Automated program repair (APR) has great potential to reduce the effort and time-consumption in software maintenance and becomes a hot topic in software engineering recently with many approaches being proposed. Multi-location program repair has always been a challenge in this field since its complexity in logic and structure. While some approaches do not claim to have the features for solving multi-location bugs, they generate correct patches for these defects in practice. In this paper, we first make an observation on multi-location bugs in Defects4J and divide them into two categories (i.e., similar and relevant multi-location bugs) based on the repair actions in their patches. We then summarize the situation of multi-location bugs in Defects4J fixed by current tools. We analyze the twenty-two patches generated by current tools and propose two feasible strategies for fixing multi-location bugs, illustrating them through two detailed case studies. At last, the experimental results prove the feasibility of our methods with the repair of two bugs that have never been fixed before. By learning from successful experience in the past, this paper points out possible ways ahead for multi-location program repair.