The research on the reform paradigm of crane hydraulic system experimental instruction based on AMESim seeks to investigate how to optimize the conventional crane hydraulic experimental teaching through simulation tec...
详细信息
Emotions play a vital role in students' academic performance and overall well-being. However, detecting and understanding students' emotions remains a challenging task for educators. This study proposes a faci...
详细信息
This research aims to predict handwritten digits using the innovative Novel Progressive VGG19 algorithm, contrasting the SVM performance. The Novel Progressive VGG19 is applied to a sample size of 74 sets, while the S...
详细信息
This paper describes a system that aims to provide course instructors with the capability to create and conduct formative and summative peer code review exercises for their students. It also aims to improve student en...
详细信息
This paper proposes the Hierarchical learning Backtracking Search Algorithm (HLBSA) for global optimization. HLBSA divides the population into three subpopulations: elite, ordinary, and inferior groups. In comparison ...
详细信息
Software Technology Foundation is an important professional foundation course in engineering colleges. To resolve the problems of the traditional teaching mode, which relies on lectures by teachers and hard to improve...
详细信息
BdSL (Bangla Sign Language) is a language in its own right, with its own grammar, syntax, and rules for word order. There are 36 alphabets in the Bangla sign language (BdSL), some of which have pretty similar characte...
详细信息
Solar panels serve as an efficient and sustainable energy source, enabling the widespread adoption of solar power as a clean alternative to conventional fuels. However, defects arising from manufacturing, transportati...
详细信息
The proliferation of online reviews has significantly impacted consumer decision-making processes, making the detection of fake reviews crucial for maintaining trust in e-commerce platforms. This study presents a nove...
详细信息
Cloud application security initiates with the analysis of security requirements in DevOps. This involves gathering, managing, and tracking requirements within integrated issue-tracking systems found in repositories li...
详细信息
ISBN:
(纸本)9798350339826
Cloud application security initiates with the analysis of security requirements in DevOps. This involves gathering, managing, and tracking requirements within integrated issue-tracking systems found in repositories like GitHub. DevOps offers advantages in cloud app development, such as accelerated deployment, improved collaboration, and enhanced reliability. In DevOps, while many security verification tools are automated, security requirements analysis often relies on manual procedures. User feedback plays a pivotal role in shaping cloud application requirements, and the industry actively seeks automation solutions to expedite development. Prior research has demonstrated the limited performance of conventional NLP models trained on established datasets, such as PROMISE, when employed in the context of GitHub Issues. Recent studies have explored the integration of deep learning, particularly leveraging modern large language models and transfer learning architectures, to address requirements engineering challenges. However, a significant issue persists - the transferability of these models. While these models excel when applied to datasets similar to those they were trained on, their performance often drastically falls when dealing with external domains. In our paper, we introduce an automated method for classifying requirements within issue trackers. This method utilizes a novel dataset comprising 12,000 security and non-security issues collected from open GitHub repositories. We employed a SmallBERT-based model for training and conducted a series of experiments. Our research reaffirms the challenge related to the transferability of NLP models. Simultaneously, our model yields highly promising results when applied to GitHub Issues, even in challenging scenarios involving issues from projects that were not part of the training dataset and structured requirements texts from the PROMISE dataset. In summary, our approach significantly contributes to enhancing DevOps prac
暂无评论