Selected work

Representative Papers

Stylometry Analysis of Human and Machine Text for Academic Integrity

Jan 03, 2026arXiv.org

This work proposes a unified natural language processing framework to address key challenges in academic integrity, including plagiarism, content fabrication, and authorship verification. The framework integrates four core stylometric tasks: classification of human- versus machine-generated text, distinction between single- and multi-author documents, detection of authorship changes within multi-author texts, and identification of contributing authors in collaborative writing. The study introduces and publicly releases the first academic text dataset generated using Gemini under two distinct instruction settings—standard and strict—and systematically evaluates how prompting strategies affect detection performance. Experimental results demonstrate that texts produced under strict instructions are significantly more adversarial, thereby increasing the difficulty of accurate identification. The code and dataset are made openly available, establishing a new benchmark for research on academic integrity.

1 citationsRead paper

Tracking the Unseen: An Occlusion-Robust Framework for Target Tracking Under Full and Long-Term Occlusion

Sep 15, 2026

该研究针对目标在完全和长期遮挡下难以持续跟踪的问题，提出了一种结合YOLOv11n、卡尔曼滤波器及遮挡感知重识别技术的鲁棒框架，有效提高了目标身份保持与轨迹连续性。

0 citationsRead paper

The Perception and Impact of Non-inclusive Language in Software Artifacts

Jul 01, 2026

This study addresses the lack of empirical evidence on how non-inclusive language in software artifacts affects developers’ sense of belonging, collaboration, and psychological well-being. Conducting the first large-scale online survey of 1,212 open-source contributors, the research systematically quantifies perceptual differences and psychological impacts associated with terms such as “whitelist/blacklist” across gender and geographic groups. Findings reveal that women and non-binary developers are significantly more likely to perceive these terms as exclusionary, which in turn correlates with markedly reduced feelings of belonging and overall well-being. By providing robust empirical support for the tangible harms of non-inclusive terminology, this work offers critical evidence to inform and motivate inclusive language practices within technical communities.

0 citationsRead paper

Tamaththul3D: High-Fidelity 3D Saudi Sign Language Avatars from Monocular Video

May 06, 2026

This work addresses the lack of high-quality 3D annotations and dedicated reconstruction methods for Arabic sign language communities by proposing the first high-fidelity 3D avatar generation framework tailored to Saudi Sign Language. Built upon the newly introduced Ishara-500 dataset featuring high-quality SMPL-X annotations, the framework introduces Tamaththul3D—a specialized pipeline that integrates SMPLer-X for full-body pose estimation, WiLoR for refined hand reconstruction (including automatic localization and mirroring), and MediaPipe-based 2D pose supervision. Through wrist alignment via kinematic chains and a hybrid swing-twist decomposition, the method optimizes gesture articulation. Experiments demonstrate that, while preserving body pose accuracy, the approach improves hand reconstruction accuracy by up to 32% over existing methods, achieving the first complete high-fidelity 3D avatar reconstruction for Arabic sign language.

0 citationsRead paper

Evaluating Large Language Models for Code Translation: Effects of Prompt Language and Prompt Design

Sep 16, 2025

This study systematically evaluates large language models (LLMs) for cross-language code translation among C++, Java, Python, and C#, addressing a gap in empirical, multi-language, multi-prompt comparative analysis. Methodologically, it introduces the first comparison of English versus Arabic prompt languages, combines concise instructions with detailed specifications as two prompt styles, employs direction-aware paired evaluation, and quantifies performance using BLEU and CodeBLEU metrics, with TransCoder as a traditional baseline. Key contributions include: (1) demonstrating that all evaluated LLMs significantly outperform TransCoder; (2) revealing dual advantages of detailed prompts and English-language prompts—specifically, English prompts improve CodeBLEU by 13–15%; and (3) proposing evidence-based prompt engineering guidelines tailored to code translation, supporting practical software migration and cross-language interoperability. The findings provide rigorous, actionable insights for leveraging LLMs in real-world multilingual software engineering tasks.

0 citationsRead paper

Recent publications

Latest Papers

Tracking the Unseen: An Occlusion-Robust Framework for Target Tracking Under Full and Long-Term Occlusion

Sep 15, 2026

该研究针对目标在完全和长期遮挡下难以持续跟踪的问题，提出了一种结合YOLOv11n、卡尔曼滤波器及遮挡感知重识别技术的鲁棒框架，有效提高了目标身份保持与轨迹连续性。

0 citationsRead paper

The Perception and Impact of Non-inclusive Language in Software Artifacts

Jul 01, 2026

This study addresses the lack of empirical evidence on how non-inclusive language in software artifacts affects developers’ sense of belonging, collaboration, and psychological well-being. Conducting the first large-scale online survey of 1,212 open-source contributors, the research systematically quantifies perceptual differences and psychological impacts associated with terms such as “whitelist/blacklist” across gender and geographic groups. Findings reveal that women and non-binary developers are significantly more likely to perceive these terms as exclusionary, which in turn correlates with markedly reduced feelings of belonging and overall well-being. By providing robust empirical support for the tangible harms of non-inclusive terminology, this work offers critical evidence to inform and motivate inclusive language practices within technical communities.

0 citationsRead paper

Tamaththul3D: High-Fidelity 3D Saudi Sign Language Avatars from Monocular Video

May 06, 2026

This work addresses the lack of high-quality 3D annotations and dedicated reconstruction methods for Arabic sign language communities by proposing the first high-fidelity 3D avatar generation framework tailored to Saudi Sign Language. Built upon the newly introduced Ishara-500 dataset featuring high-quality SMPL-X annotations, the framework introduces Tamaththul3D—a specialized pipeline that integrates SMPLer-X for full-body pose estimation, WiLoR for refined hand reconstruction (including automatic localization and mirroring), and MediaPipe-based 2D pose supervision. Through wrist alignment via kinematic chains and a hybrid swing-twist decomposition, the method optimizes gesture articulation. Experiments demonstrate that, while preserving body pose accuracy, the approach improves hand reconstruction accuracy by up to 32% over existing methods, achieving the first complete high-fidelity 3D avatar reconstruction for Arabic sign language.

0 citationsRead paper

Stylometry Analysis of Human and Machine Text for Academic Integrity

Jan 03, 2026arXiv.org

This work proposes a unified natural language processing framework to address key challenges in academic integrity, including plagiarism, content fabrication, and authorship verification. The framework integrates four core stylometric tasks: classification of human- versus machine-generated text, distinction between single- and multi-author documents, detection of authorship changes within multi-author texts, and identification of contributing authors in collaborative writing. The study introduces and publicly releases the first academic text dataset generated using Gemini under two distinct instruction settings—standard and strict—and systematically evaluates how prompting strategies affect detection performance. Experimental results demonstrate that texts produced under strict instructions are significantly more adversarial, thereby increasing the difficulty of accurate identification. The code and dataset are made openly available, establishing a new benchmark for research on academic integrity.

1 citationsRead paper

Evaluating Large Language Models for Code Translation: Effects of Prompt Language and Prompt Design

Sep 16, 2025

This study systematically evaluates large language models (LLMs) for cross-language code translation among C++, Java, Python, and C#, addressing a gap in empirical, multi-language, multi-prompt comparative analysis. Methodologically, it introduces the first comparison of English versus Arabic prompt languages, combines concise instructions with detailed specifications as two prompt styles, employs direction-aware paired evaluation, and quantifies performance using BLEU and CodeBLEU metrics, with TransCoder as a traditional baseline. Key contributions include: (1) demonstrating that all evaluated LLMs significantly outperform TransCoder; (2) revealing dual advantages of detailed prompts and English-language prompts—specifically, English prompts improve CodeBLEU by 13–15%; and (3) proposing evidence-based prompt engineering guidelines tailored to code translation, supporting practical software migration and cross-language interoperability. The findings provide rigorous, actionable insights for leveraging LLMs in real-world multilingual software engineering tasks.

0 citationsRead paper