ToxiSpanSE: An Explainable Toxicity Detection in Code Review Comments

Sarker, Jaydeb; Sultana, Sayma; Wilson, Steven R.; Bosu, Amiangshu

doi:10.1109/esem56168.2023.10304855

Cited by 5 publications

(2 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…When considering toxicity as a general term, we simply need to take into account its characteristics and assess each person's ability to address them [12,14,15,. • Software Engineering (SE): The term is defined similarly but with more emphasis on its occurrence in conversations and distinctions such as name-calling, flirtation, references to sexual activities, microaggressions, trolling, flaming, arrogance, and stereotyping [87][88][89][90][91]. As it pertains to a professional environment, differentiating characteristics emerge, and specific language nuances, such as the varied interpretations of terms like kill are considered.…”

Section: Systematic Review: Outcomesmentioning

confidence: 99%

A Systematic Review of Toxicity in Large Language Models: Definitions, Datasets, Detectors, Detoxification Methods and Challenges

Villate-Castillo,

Ser,

Urquijo

2024

Preprint

View full text Add to dashboard Cite

The emergence of the transformer architecture has ushered in a new era of possibilities, showcasing remarkable capabilities in generative tasks exemplified by models like GPT4o, Claude 3, and Llama 3. However, these advancements come with a caveat: predominantly trained on data gleaned from social media platforms, these systems inadvertently perpetuate societal biases and toxicity. Recognizing the paramount importance of AI Safety and Alignment, our study embarks on a thorough exploration through a comprehensive literature review focused on toxic language. Delving into various definitions, detection methodologies, and mitigation strategies, we aim to shed light on the complexities of this issue. While our focus primarily centres on transformer-based architectures, we also acknowledge and incorporate existing research within the realm of deep learning. Through our investigation, we uncover a multitude of challenges inherent in toxicity mitigation and detection models. These challenges range from inherent biases and generalization issues to the necessity for standardized definitions of toxic language and the quality assurance of dataset annotations. Furthermore, we emphasize the significance of transparent annotation processes, resolution of annotation disagreements, and the enhancement of Large Language Models (LLMs) robustness. Additionally, we advocate for the creation of standardized benchmarks to gauge the effectiveness of toxicity mitigation and detection methods. Addressing these challenges is not just imperative, but pivotal in advancing the development of safer and more ethically aligned AI systems.

show abstract

Section: Systematic Review: Outcomesmentioning

confidence: 99%

A Systematic Review of Toxicity in Large Language Models: Definitions, Datasets, Detectors, Detoxification Methods and Challenges

Villate-Castillo,

Ser,

Urquijo

2024

Preprint

View full text Add to dashboard Cite

show abstract

“…In this thriving online ecosystem, user reviews serve as the primary source of information for users to understand the practical aspects of various applications [2,3]. However, a significant amount of inappropriate content unrelated to the application itself is being disseminated through these online reviews, including descriptions and links related to pornography, gambling, and other activities [4,5]. These posts not only impact users' browsing experience but also require them to spend more time searching for information that is beneficial to them [6].…”

Section: Introductionmentioning

confidence: 99%

Detection System Based on Text Adversarial and Multi-Information Fusion for Inappropriate Comments in Mobile Application Reviews

Yu,

Jia,

Hong

2024

Electronics

View full text Add to dashboard Cite

With the rapid development of mobile application technology, the content and forms of comments disseminated on the internet are becoming increasingly complex. Various comments serve as users’ firsthand reference materials for understanding the application. However, some comments contain a significant amount of inappropriate content unrelated to the app itself, such as gambling, loans, pornography, and game account recharging, seriously impacting the user experience. Therefore, this article aims to assist users in filtering out irrelevant and inappropriate messages, enabling them to quickly obtain useful and relevant information. This study focuses on analyzing actual comments on various Chinese apps on the Apple App Store. However, these irrelevant comments exhibit a certain degree of concealment, sparsity, and complexity, which increases the difficulty of detection. Additionally, due to language differences, the existing English research methods exhibit relatively poor adaptability to Chinese textual data. To overcome these challenges, this paper proposes a research method named “blend net”, which combines text adversarial and multi-information fusion detection to enhance the overall performance of the system. The experimental results demonstrate that the method proposed in this paper achieves precision and recall rates both exceeding 98%, representing an improvement of at least 2% compared to existing methods.

show abstract

Exploring ChatGPT for identifying sexism in the communication of software developers

Sultana,

Begum Kali

2024

Proceedings of the 17th International Conference on PErvasive Technologies Related to Assistive Environments

View full text Add to dashboard Cite

ToxiSpanSE: An Explainable Toxicity Detection in Code Review Comments

Cited by 5 publications

References 44 publications

A Systematic Review of Toxicity in Large Language Models: Definitions, Datasets, Detectors, Detoxification Methods and Challenges

A Systematic Review of Toxicity in Large Language Models: Definitions, Datasets, Detectors, Detoxification Methods and Challenges

Detection System Based on Text Adversarial and Multi-Information Fusion for Inappropriate Comments in Mobile Application Reviews

Exploring ChatGPT for identifying sexism in the communication of software developers

Contact Info

Product

Resources

About