You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Software comments are critical for human understanding of software, and as such many comment generation techniques have been proposed. However, we find that a systematic evaluation of the factual accuracy of generated comments is rare; only subjective accuracy labels have been given. Evaluating comments generated by three Large Language Models (LLMs), we find that even for the best-performing LLM, roughly a fifth of its comments contained demonstrably inaccurate statements. While it seems code-comment consistency detection techniques should be able to detect inaccurate comments, we perform experiments demonstrating they have no statistically significant relationship with comment accuracy, underscoring the substantial difficulty of this problem. To tackle this, we propose the concept of document testing, in which a document is verified by using an LLM to generate tests based on the document, running those tests, and observing whether they pass or fail. Furthermore, we implement our concept to verify Java comments. Experiments demonstrate that our approach has a robust statistical relationship with comment accuracy, making headway into a problem where prior techniques failed. Qualitative evaluation also reveals the promise of our approach in gaining developer trust, while highlighting the limitations of our current implementation.
Authors: Zeyao Ma, Bohan Zhang, Jing Zhang, Jifan Yu, Xiaokang Zhang, Xiaohan Zhang, Sijia Luo, Xi Wang, Jie Tang
Tags: LLM
We introduce SpreadsheetBench, a challenging spreadsheet manipulation benchmark exclusively derived from real-world scenarios, designed to immerse current large language models (LLMs) in the actual workflow of spreadsheet users. Unlike existing benchmarks that rely on synthesized queries and simplified spreadsheet files, SpreadsheetBench is built from 912 real questions gathered from online Excel forums, which reflect the intricate needs of users. The associated spreadsheets from the forums contain a variety of tabular data such as multiple tables, non-standard relational tables, and abundant non-textual elements. Furthermore, we propose a more reliable evaluation metric akin to online judge platforms, where multiple spreadsheet files are created as test cases for each instruction, ensuring the evaluation of robust solutions capable of handling spreadsheets with varying values. Our comprehensive evaluation of various LLMs under both single-round and multi-round inference settings reveals a substantial gap between the state-of-the-art (SOTA) models and human performance, highlighting the benchmark's difficulty.
Authors: Santiago Berrezueta-Guzman, Mohanad Kandil, Mar'ia-Luisa Mart'in-Ruiz, Iv'an Pau-de-la-Cruz, Stephan Krusche
Tags: LLM
Attention Deficit Hyperactivity Disorder (ADHD) is a neurodevelopmental condition characterized by inattention, hyperactivity, and impulsivity, which can significantly impact an individual's daily functioning and quality of life. Occupational therapy plays a crucial role in managing ADHD by fostering the development of skills needed for daily living and enhancing an individual's ability to participate fully in school, home, and social situations. Recent studies highlight the potential of integrating Large Language Models (LLMs) like ChatGPT and Socially Assistive Robots (SAR) to improve psychological treatments. This integration aims to overcome existing limitations in mental health therapy by providing tailored support and adapting to the unique needs of this sensitive group. However, there remains a significant gap in research exploring the combined use of these advanced technologies in ADHD therapy, suggesting an opportunity for novel therapeutic approaches.
Thus, we integrated two advanced language models, ChatGPT-4 Turbo and Claude-3 Opus, into a robotic assistant to explore how well each model performs in robot-assisted interactions. Additionally, we have compared their performance in a simulated therapy scenario to gauge their effectiveness against a clinically validated customized model. The results of this study show that ChatGPT-4 Turbo excelled in performance and responsiveness, making it suitable for time-sensitive applications. Claude-3 Opus, on the other hand, showed strengths in understanding, coherence, and ethical considerations, prioritizing safe and engaging interactions. Both models demonstrated innovation and adaptability, but ChatGPT-4 Turbo offered greater ease of integration and broader language support. The selection between them hinges on the specific demands of ADHD therapy.
Recent research in Needle-in-a-Haystack (NIAH) benchmarks has explored the capabilities of Large Language Models (LLMs) in retrieving contextual information from large text documents. However, as LLMs become increasingly integrated into software development processes, it is crucial to evaluate their performance in code-based environments. As LLMs are further developed for program synthesis, we need to ensure that LLMs can understand syntax and write syntactically correct code. As a step in ensuring LLMs understand syntax, LLMs can be evaluated in their ability to find and detect syntax bugs. Our benchmark, Bug In The Code Stack (BICS), is designed to assess the ability of LLMs to identify simple syntax bugs within large source code. Our findings reveal three key insights: (1) code-based environments pose significantly more challenge compared to text-based environments for retrieval tasks, (2) there is a substantial performance disparity among different models, and (3) there is a notable correlation between longer context lengths and performance degradation, though the extent of this degradation varies between models.
Authors: Lingzhe Zhang, Tong Jia, Mengxi Jia, Yifan Wu, Aiwei Liu, Yong Yang, Zhonghai Wu, Xuming Hu, Philip S. Yu, Ying Li
Tags: LLM
As software systems grow increasingly intricate, Artificial Intelligence for IT Operations (AIOps) methods have been widely used in software system failure management to ensure the high availability and reliability of large-scale distributed software systems. However, these methods still face several challenges, such as lack of cross-platform generality and cross-task flexibility. Fortunately, recent advancements in large language models (LLMs) can significantly address these challenges, and many approaches have already been proposed to explore this field. However, there is currently no comprehensive survey that discusses the differences between LLM-based AIOps and traditional AIOps methods. Therefore, this paper presents a comprehensive survey of AIOps technology for failure management in the LLM era. It includes a detailed definition of AIOps tasks for failure management, the data sources for AIOps, and the LLM-based approaches adopted for AIOps. Additionally, this survey explores the AIOps subtasks, the specific LLM-based approaches suitable for different AIOps subtasks, and the challenges and future directions of the domain, aiming to further its development and application.
Authors: Lucas Franke, Huayu Liang, Sahar Farzanehpour, Aaron Brantly, James C. Davis, Chris Brown
Background: Governments worldwide are considering data privacy regulations. These laws, e.g. the European Union's General Data Protection Regulation (GDPR), require software developers to meet privacy-related requirements when interacting with users' data. Prior research describes the impact of such laws on software development, but only for commercial software. Open-source software is commonly integrated into regulated software, and thus must be engineered or adapted for compliance. We do not know how such laws impact open-source software development.
Aims: To understand how data privacy laws affect open-source software development. We studied the European Union's GDPR, the most prominent such law. We investigated how GDPR compliance activities influence OSS developer activity (RQ1), how OSS developers perceive fulfilling GDPR requirements (RQ2), the most challenging GDPR requirements to implement (RQ3), and how OSS developers assess GDPR compliance (RQ4).
Method: We distributed an online survey to explore perceptions of GDPR implementations from open-source developers (N=56). We further conducted a repository mining study to analyze development metrics on pull requests (N=31462) submitted to open-source GitHub repositories.
Results: GDPR policies complicate open-source development processes and introduce challenges for developers, primarily regarding the management of users' data, implementation costs and time, and assessments of compliance. Moreover, we observed negative perceptions of GDPR from open-source developers and significant increases in development activity, in particular metrics related to coding and reviewing activity, on GitHub pull requests related to GDPR compliance.
Conclusions: Our findings motivate policy-related resources and automated tools to support data privacy regulation implementation and compliance efforts in open-source software.
Authors: Rafael S'a Menezes, Edoardo Manino, Fedor Shmarov, Mohannad Aldughaim, Rosiane de Freitas, Lucas C. Cordeiro
Bounded Model Checking (BMC) is a widely used software verification technique. Despite its successes, the technique has several limiting factors, from state-space explosion to lack of completeness. Over the years, interval analysis has repeatedly been proposed as a partial solution to these limitations. In this work, we evaluate whether the computational cost of interval analysis yields significant enough improvements in BMC's performance to justify its use. In more detail, we quantify the benefits of interval analysis on two benchmarks: the Intel Core Power Management firmware and 9537 programs in the ReachSafety category of the International Competition on Software Verification. Our results show that interval analysis is essential in solving 203 unique benchmarks.
Authors: Willian D. F. Mendon\c{c}a, Wesley K. G. Assun\c{c}~ao, Silvia R. Vergilio
Testing Highly Configurable Systems (HCSs) is a challenging task, especially in an evolution scenario where features are added, changed, or removed, which hampers test case selection and prioritization. Existing work is usually based on the variability model, which is not always available or updated. Yet, the few existing approaches rely on links between test cases and changed files (or lines of code), not considering how features are implemented, usually spread over several and unchanged files. To overcome these limitations, we introduce FeaTestSelPrio, a feature-oriented test case selection and prioritization approach for HCSs. The approach links test cases to feature implementations, using HCS pre-processor directives, to select test cases based on features affected by changes in each commit. After, the test cases are prioritized according to the number of features they cover. Our approach selects a greater number of tests and takes longer to execute than a changed-file-oriented approach, used as baseline, but FeaTestSelPrio performs better regarding detected failures. By adding the approach execution time to the execution time of the selected test cases, we reached a reduction of $\approx$50%, in comparison with retest-all. The prioritization step allows reducing the average test budget in 86% of the failed commits.
Software practitioners discuss problems at work with peers, in-person and online. These discussions can be technical (e.g., how to fix a bug?) and social (e.g., how to assign work fairly?). While there is a growing body of knowledge exploring fairness problems and solutions in the human and social factors of software engineering, most focus has been on specific problems. This study provides fairness discussions by software practitioners on Stack Exchange sites. We present an exploratory study presenting the fairness experience of software practitioners and fairness expectations in software teams. We also want to identify the fairness aspects software practitioners talk about the most. For example, do they care more about fairness in income or how they are treated in the workplace?
Our investigation of fairness discussions on eight Stack Exchange sites resulted in a list of 136 posts (28 questions and 108 answers) manually curated from 4,178 candidate posts. The study reveals that the majority of fairness discussions (24 posts) revolve around the topic of income suggesting that many software practitioners are highly interested in matters related to their pay and how it is fairly distributed. Further, we noted that while not discussed as often, discussions on fairness in recruitment tend to receive the highest number of views and scores. Interestingly, the study shows that unfairness experiences extend beyond the protected attributes. In this study, only 25 out of 136 posts mention protected attributes, with gender mainly being discussed.
The text was updated successfully, but these errors were encountered:
cs.SE updates on arXiv.org, Mon, 24 Jun 2024 04:00:07 +0000
Identifying Inaccurate Descriptions in LLM-generated Code Comments via Test Execution [pdf]
Authors: Sungmin Kang, Louis Milliken, Shin Yoo
Tags:
LLMSoftware comments are critical for human understanding of software, and as such many comment generation techniques have been proposed. However, we find that a systematic evaluation of the factual accuracy of generated comments is rare; only subjective accuracy labels have been given. Evaluating comments generated by three Large Language Models (LLMs), we find that even for the best-performing LLM, roughly a fifth of its comments contained demonstrably inaccurate statements. While it seems code-comment consistency detection techniques should be able to detect inaccurate comments, we perform experiments demonstrating they have no statistically significant relationship with comment accuracy, underscoring the substantial difficulty of this problem. To tackle this, we propose the concept of document testing, in which a document is verified by using an LLM to generate tests based on the document, running those tests, and observing whether they pass or fail. Furthermore, we implement our concept to verify Java comments. Experiments demonstrate that our approach has a robust statistical relationship with comment accuracy, making headway into a problem where prior techniques failed. Qualitative evaluation also reveals the promise of our approach in gaining developer trust, while highlighting the limitations of our current implementation.
SpreadsheetBench: Towards Challenging Real World Spreadsheet Manipulation [pdf]
Authors: Zeyao Ma, Bohan Zhang, Jing Zhang, Jifan Yu, Xiaokang Zhang, Xiaohan Zhang, Sijia Luo, Xi Wang, Jie Tang
Tags:
LLMWe introduce SpreadsheetBench, a challenging spreadsheet manipulation benchmark exclusively derived from real-world scenarios, designed to immerse current large language models (LLMs) in the actual workflow of spreadsheet users. Unlike existing benchmarks that rely on synthesized queries and simplified spreadsheet files, SpreadsheetBench is built from 912 real questions gathered from online Excel forums, which reflect the intricate needs of users. The associated spreadsheets from the forums contain a variety of tabular data such as multiple tables, non-standard relational tables, and abundant non-textual elements. Furthermore, we propose a more reliable evaluation metric akin to online judge platforms, where multiple spreadsheet files are created as test cases for each instruction, ensuring the evaluation of robust solutions capable of handling spreadsheets with varying values. Our comprehensive evaluation of various LLMs under both single-round and multi-round inference settings reveals a substantial gap between the state-of-the-art (SOTA) models and human performance, highlighting the benchmark's difficulty.
Exploring the Efficacy of Robotic Assistants with ChatGPT and Claude in Enhancing ADHD Therapy: Innovating Treatment Paradigms [pdf]
Authors: Santiago Berrezueta-Guzman, Mohanad Kandil, Mar'ia-Luisa Mart'in-Ruiz, Iv'an Pau-de-la-Cruz, Stephan Krusche
Tags:
LLMAttention Deficit Hyperactivity Disorder (ADHD) is a neurodevelopmental condition characterized by inattention, hyperactivity, and impulsivity, which can significantly impact an individual's daily functioning and quality of life. Occupational therapy plays a crucial role in managing ADHD by fostering the development of skills needed for daily living and enhancing an individual's ability to participate fully in school, home, and social situations. Recent studies highlight the potential of integrating Large Language Models (LLMs) like ChatGPT and Socially Assistive Robots (SAR) to improve psychological treatments. This integration aims to overcome existing limitations in mental health therapy by providing tailored support and adapting to the unique needs of this sensitive group. However, there remains a significant gap in research exploring the combined use of these advanced technologies in ADHD therapy, suggesting an opportunity for novel therapeutic approaches.
Thus, we integrated two advanced language models, ChatGPT-4 Turbo and Claude-3 Opus, into a robotic assistant to explore how well each model performs in robot-assisted interactions. Additionally, we have compared their performance in a simulated therapy scenario to gauge their effectiveness against a clinically validated customized model. The results of this study show that ChatGPT-4 Turbo excelled in performance and responsiveness, making it suitable for time-sensitive applications. Claude-3 Opus, on the other hand, showed strengths in understanding, coherence, and ethical considerations, prioritizing safe and engaging interactions. Both models demonstrated innovation and adaptability, but ChatGPT-4 Turbo offered greater ease of integration and broader language support. The selection between them hinges on the specific demands of ADHD therapy.
Bug In the Code Stack: Can LLMs Find Bugs in Large Python Code Stacks [pdf]
Authors: Hokyung Lee, Sumanyu Sharma, Bing Hu
Tags:
LLMRecent research in Needle-in-a-Haystack (NIAH) benchmarks has explored the capabilities of Large Language Models (LLMs) in retrieving contextual information from large text documents. However, as LLMs become increasingly integrated into software development processes, it is crucial to evaluate their performance in code-based environments. As LLMs are further developed for program synthesis, we need to ensure that LLMs can understand syntax and write syntactically correct code. As a step in ensuring LLMs understand syntax, LLMs can be evaluated in their ability to find and detect syntax bugs. Our benchmark, Bug In The Code Stack (BICS), is designed to assess the ability of LLMs to identify simple syntax bugs within large source code. Our findings reveal three key insights: (1) code-based environments pose significantly more challenge compared to text-based environments for retrieval tasks, (2) there is a substantial performance disparity among different models, and (3) there is a notable correlation between longer context lengths and performance degradation, though the extent of this degradation varies between models.
A Survey of AIOps for Failure Management in the Era of Large Language Models [pdf]
Authors: Lingzhe Zhang, Tong Jia, Mengxi Jia, Yifan Wu, Aiwei Liu, Yong Yang, Zhonghai Wu, Xuming Hu, Philip S. Yu, Ying Li
Tags:
LLMAs software systems grow increasingly intricate, Artificial Intelligence for IT Operations (AIOps) methods have been widely used in software system failure management to ensure the high availability and reliability of large-scale distributed software systems. However, these methods still face several challenges, such as lack of cross-platform generality and cross-task flexibility. Fortunately, recent advancements in large language models (LLMs) can significantly address these challenges, and many approaches have already been proposed to explore this field. However, there is currently no comprehensive survey that discusses the differences between LLM-based AIOps and traditional AIOps methods. Therefore, this paper presents a comprehensive survey of AIOps technology for failure management in the LLM era. It includes a detailed definition of AIOps tasks for failure management, the data sources for AIOps, and the LLM-based approaches adopted for AIOps. Additionally, this survey explores the AIOps subtasks, the specific LLM-based approaches suitable for different AIOps subtasks, and the challenges and future directions of the domain, aiming to further its development and application.
An Exploratory Mixed-Methods Study on General Data Protection Regulation (GDPR) Compliance in Open-Source Software [pdf]
Authors: Lucas Franke, Huayu Liang, Sahar Farzanehpour, Aaron Brantly, James C. Davis, Chris Brown
Background: Governments worldwide are considering data privacy regulations. These laws, e.g. the European Union's General Data Protection Regulation (GDPR), require software developers to meet privacy-related requirements when interacting with users' data. Prior research describes the impact of such laws on software development, but only for commercial software. Open-source software is commonly integrated into regulated software, and thus must be engineered or adapted for compliance. We do not know how such laws impact open-source software development.
Aims: To understand how data privacy laws affect open-source software development. We studied the European Union's GDPR, the most prominent such law. We investigated how GDPR compliance activities influence OSS developer activity (RQ1), how OSS developers perceive fulfilling GDPR requirements (RQ2), the most challenging GDPR requirements to implement (RQ3), and how OSS developers assess GDPR compliance (RQ4).
Method: We distributed an online survey to explore perceptions of GDPR implementations from open-source developers (N=56). We further conducted a repository mining study to analyze development metrics on pull requests (N=31462) submitted to open-source GitHub repositories.
Results: GDPR policies complicate open-source development processes and introduce challenges for developers, primarily regarding the management of users' data, implementation costs and time, and assessments of compliance. Moreover, we observed negative perceptions of GDPR from open-source developers and significant increases in development activity, in particular metrics related to coding and reviewing activity, on GitHub pull requests related to GDPR compliance.
Conclusions: Our findings motivate policy-related resources and automated tools to support data privacy regulation implementation and compliance efforts in open-source software.
Interval Analysis in Industrial-Scale BMC Software Verifiers: A Case Study [pdf]
Authors: Rafael S'a Menezes, Edoardo Manino, Fedor Shmarov, Mohannad Aldughaim, Rosiane de Freitas, Lucas C. Cordeiro
Bounded Model Checking (BMC) is a widely used software verification technique. Despite its successes, the technique has several limiting factors, from state-space explosion to lack of completeness. Over the years, interval analysis has repeatedly been proposed as a partial solution to these limitations. In this work, we evaluate whether the computational cost of interval analysis yields significant enough improvements in BMC's performance to justify its use. In more detail, we quantify the benefits of interval analysis on two benchmarks: the Intel Core Power Management firmware and 9537 programs in the ReachSafety category of the International Competition on Software Verification. Our results show that interval analysis is essential in solving 203 unique benchmarks.
Feature-oriented Test Case Selection and Prioritization During the Evolution of Highly-Configurable Systems [pdf]
Authors: Willian D. F. Mendon\c{c}a, Wesley K. G. Assun\c{c}~ao, Silvia R. Vergilio
Testing Highly Configurable Systems (HCSs) is a challenging task, especially in an evolution scenario where features are added, changed, or removed, which hampers test case selection and prioritization. Existing work is usually based on the variability model, which is not always available or updated. Yet, the few existing approaches rely on links between test cases and changed files (or lines of code), not considering how features are implemented, usually spread over several and unchanged files. To overcome these limitations, we introduce FeaTestSelPrio, a feature-oriented test case selection and prioritization approach for HCSs. The approach links test cases to feature implementations, using HCS pre-processor directives, to select test cases based on features affected by changes in each commit. After, the test cases are prioritized according to the number of features they cover. Our approach selects a greater number of tests and takes longer to execute than a changed-file-oriented approach, used as baseline, but FeaTestSelPrio performs better regarding detected failures. By adding the approach execution time to the execution time of the selected test cases, we reached a reduction of $\approx$50%, in comparison with retest-all. The prioritization step allows reducing the average test budget in 86% of the failed commits.
Understanding Fairness in Software Engineering: Insights from Stack Exchange [pdf]
Authors: Emeralda Sesari, Federica Sarro, Ayushi Rastogi
Software practitioners discuss problems at work with peers, in-person and online. These discussions can be technical (e.g., how to fix a bug?) and social (e.g., how to assign work fairly?). While there is a growing body of knowledge exploring fairness problems and solutions in the human and social factors of software engineering, most focus has been on specific problems. This study provides fairness discussions by software practitioners on Stack Exchange sites. We present an exploratory study presenting the fairness experience of software practitioners and fairness expectations in software teams. We also want to identify the fairness aspects software practitioners talk about the most. For example, do they care more about fairness in income or how they are treated in the workplace?
Our investigation of fairness discussions on eight Stack Exchange sites resulted in a list of 136 posts (28 questions and 108 answers) manually curated from 4,178 candidate posts. The study reveals that the majority of fairness discussions (24 posts) revolve around the topic of income suggesting that many software practitioners are highly interested in matters related to their pay and how it is fairly distributed. Further, we noted that while not discussed as often, discussions on fairness in recruitment tend to receive the highest number of views and scores. Interestingly, the study shows that unfairness experiences extend beyond the protected attributes. In this study, only 25 out of 136 posts mention protected attributes, with gender mainly being discussed.
The text was updated successfully, but these errors were encountered: