Due to the widespread integration of embedded devices in security-critical and safety-critical systems, an effective attack poses a significant threat to these vital systems. The compilation and execution characterist...
详细信息
software effort estimation (SEE) in Agile software Development (ASD) has been a longstanding challenge for software engineers. The traditional approach estimates effort scores known as Story Point (SP) to measure user...
详细信息
Large Language Models (LLMs) are increasingly utilized for softwareengineering tasks, including code generation. While prior research has primarily focused on code completion, there remains a gap in understanding LLM...
详细信息
software flaws pose a serious risk to the dependability and security of computer systems and applications in a variety of industries. Because they take a long time and are prone to errors, traditional techniques of fi...
详细信息
Cataract, the leading cause of global blindness, represents a focal concern within the field of blindness prevention. Its diagnosis primarily relies on the observation of lens opacification under slit-lamp examination...
详细信息
software operates various essential systems and devices in today's world. Many companies build systems with varying sizes and functions in order to offer higher-quality software more quickly. The goal of this dive...
详细信息
Context and Motivation: Rapid advancements in Artificial Intelligence (AI) have significantly influenced requirements engineering (RE) practices. Problem: While many recent secondary studies have explored AI's rol...
详细信息
ISBN:
(纸本)9783031573262;9783031573279
Context and Motivation: Rapid advancements in Artificial Intelligence (AI) have significantly influenced requirements engineering (RE) practices. Problem: While many recent secondary studies have explored AI's role in RE, a thorough understanding of the use of AI for RE (AI4RE) and its inherent challenges remains in its early stages. Principal Ideas: To fill this knowledge gap, we conducted a tertiary review on understanding how AI assists RE practices. Contribution: We analyzed 28 secondary studies from 2017 to September 2023 about using AI in RE tasks such as elicitation, classification, analysis, specification, management, and tracing. Our study reveals a trend of combining natural language process techniques with machine learning models like Latent Dirichlet Allocation (LDA) and Naive Bayes, and a surge in using large language models (LLMs) for RE. The study also identified challenges of AI4RE related to ambiguity, language, data, algorithm, and evaluation. The study gives topics for future research, particularly for researchers who want to start new research in this field.
softwareengineering aims to effectively translate stakeholders' requirements into executable code to fulfill their needs. Traceability from natural language use case requirements to classes in a UML class diagram...
详细信息
ISBN:
(纸本)9783031758713;9783031758720
softwareengineering aims to effectively translate stakeholders' requirements into executable code to fulfill their needs. Traceability from natural language use case requirements to classes in a UML class diagram, subsequently translated into code implementation, is essential in systems development and maintenance. Tasks such as assessing the impact of changes and enhancing software reusability require a clear link between these requirements and their software implementation. However, establishing such links manually across extensive codebases is prohibitively challenging. Requirements, typically articulated in natural language, embody semantics that clarify the purpose of the codebase. Conventional traceability methods, relying on textual similarities between requirements and code, often suffer from low precision due to the semantic gap between high-level natural language requirements and the syntactic nature of code. The advent of Large Language Models (LLMs) provides new methods to address this challenge through their advanced capability to interpret both natural language and code syntax. Furthermore, representing code as a knowledge graph facilitates the use of graph structural information to enhance traceability links. This paper introduces an LLM-supported retrieval augmented generation approach for enhancing requirements traceability to the class diagram of the code, incorporating keyword, vector, and graph indexing techniques, and their integrated application. We present a comparative analysis against conventional methods and among different indexing strategies and parameterizations on the performance. Our results demonstrate how this methodology significantly improves the efficiency and accuracy of establishing traceability links in software development processes.
Integrating state-of-the-art robotics algorithms into a working system remains a challenge, especially for largescale deployments. One promising way to address such issues is to apply to robotics the massive cloud and...
详细信息
ISBN:
(纸本)9798350358513;9798350358520
Integrating state-of-the-art robotics algorithms into a working system remains a challenge, especially for largescale deployments. One promising way to address such issues is to apply to robotics the massive cloud and edge computing that already powers AI research from training to deployment. In this paper, we extend the KubeROS platform to serve as a one-stop solution aimed at facilitating robotic development cycles from research to deployment. The closed-loop workflow maintains a single software stack containing many microserviceoriented and containerized software modules across all stages. In addition to the architectural adaptation, we develop a new subsystem called BatchJob to leverage cloud/edge to efficiently run experiments at scale and evaluate the system with statistical significance. Finally, our approach enables a seamless transition to production and closes the cycle for continuous improvement. To demonstrate its applicability and scalability, we choose mobile robot navigation as an example and perform experiments on three levels: comparison of localization approaches with benchmark datasets, testing and evaluation with closedloop simulation, and evaluation on a real robot. Compared to sequential execution on a single machine, our approach accelerates experiments by at least a factor of 30 on a 9node edge cluster. Our platform and code to reproduce the experiments are available on Github.
Reading, adapting, and maintaining complex software can be a daunting task. We might need to refactor it to streamline the process and make the code cleaner and self-explanatory. Traditional refactoring tools guide de...
详细信息
暂无评论