Blekinge Institute of Technology
Not a member yet
13576 research outputs found
Sort by
Natural Language Processing in Software Engineering : A Systematic Literature Review
No full textContext: Software engineering (SE) artifacts and documents, such as requirements specifications, user stories, test cases, and concepts of operations (ConOps), are typically written in natural language, making their manipulation challenging. Natural Language Processing (NLP) is a viable solution for managing these tasks. Objective: To conduct a systematic literature review to explore the current use of NLP in SE artifacts and tasks, supplementedby a tertiary study focusing on the emerging role of Large Language Models (LLMs) in software engineering re-search. Method: We searched digital libraries for relevant papers and applied inclusion and exclusion criteria to filter the primary studies. We then analyzed NLP techniques applied to SE documents and examined their usage in this context. Our research methodology followed Kitchenham and Charters’ guidelines. Additionally, we conducted a tertiary study to synthesize findings from existing systematic literature reviews and surveys specifically addressing LLMs in software engineering. Results: We selected 60 primary studies to identify the most common methods for NLP pipelines, feature extraction, language models, and machine learning algorithms used in SE. We also assessed the purposes of these methods, their benefits for SE, their difficulty, and their contribution to SE advancement. The tertiary study revealed a rapid proliferation of LLM-focused research, with comprehensive reviews documenting exponential growth in publications and widespread adoption across diverse SE tasks. Conclusion: Requirements are the most frequently addressed artifacts using NLP techniques, with preprocessing and part-of-speech (POS) tagging being widely used. There is a notable increase in the use of large language models for various SE tasks, such as requirements elicitation, source code generation, bug fixing, and software testing. The tertiary study confirms that LLMs represent a pivotal shift in the research landscape, warranting dedicated investigation to understand their transformative impact on NLP applications in software engineering
Roadmap to enable sustainable and circular designs in collaborative automotive ecosystems
No full textA roadmap for advancing sustainable and circular designs within the automotive industry is proposed. The emphasis is on the critical role of collaborative ecosystems following the increased transparency and traceability underway in regulations. Emerging Digital Product Passports are central means in Europe’s Green Deal and expects to drive transformation of practices in the automotive ecosystem. The study, conducted by researchers in collaboration with a global truck manufacturer, identifies key areas for action, including data quality, stakeholder value, and communication strategies, to facilitate the circular and sustainable transformation. The vision and actions proposed were refined in workshops with automotive suppliers and service providers. By addressing these challenges, the automotive industry can leverage from data accessibility and accelerate its shift towards sustainability.Produktutveckla med hänsyn till Digital Product Passport - DIP
Effective system-of-systems simulation in a VUCA world : lessons learned for design decision-makers
No full textToday, Manufacturing companies are adopting a servitization strategy and Product-Service System model to enhance value and remain competitive. Often, this transition also means to embrace a System-of-Systems (SoS) perspective. Concurrently, companies face challenges with volatile, uncertain, complex, and ambiguous (VUCA) environments. One way to tackle VUCA is to utilize simulation modeling. However, developing SoS simulations can be complex and cumbersome. This paper extracts lessons learned from six case studies to identify effective and ineffective practices in developing simulation models. The analysis has led to nine design principles for more effective simulation modeling. Furthermore, the paper explores simulation techniques for modeling SoS and discusses effective VUCA management. Finally, the paper proposes four future research directions to advance SoS simulation research
Psykisk hälsa i samband med endometrios – En utmaning för kvinnors psykiska hälsa
No full textBakgrund: Endometrios är en kronisk sjukdom där livmodersliknande vävnad växer utanför livmodern och orsakar smärta, inflammation och ibland fertilitetsproblem. Sjukdomen påverkar både fysisk och psykisk hälsa, där långvarig smärta och begränsningar i vardagen kan leda till påverkan på den psykiska hälsan. Sjuksköterskans roll är att ge personcentrerad vård genom att lindra smärta, erbjuda information, stöd och förståelse. Syfte: Syftet med studien var att beskriva upplevelsen av hur psykisk hälsa påverkas hos kvinnor med endometrios. Metod: Studien har genomförts igenom en allmän litteraturöversikt av kvantitativ och kvalitativ forskning. Databaserna som användes för att samla in data var Cinahl och PubMed. Kvalitetsgranskningen gjordes genom granskningsmallarna enligt Friberg. Analys gjordes genom fyra steg. Etiska övervägande hade författarna under hela arbetets gång. Resultat: Resultatet bestod av fyra kategorier begränsningar av det sociala umgänget, begränsningar av intima relationer, behov av stöd och betydelsen av att acceptera sjukdomen. Slutsats: Kvinnors upplevelser av endometrios visar att sjukdomen påverkar den psykiska hälsan. Kvinnor med endometrios upplever sociala begränsningar, smärta och bristande förståelse från omgivningen. Samtidigt framkommer det att stöd, kunskap och strategier för acceptans kan stärka känslan av kontroll och bidra till en bättre psykisk hälsa
The Impact of AI-Driven Code Review on Developer Productivity and Software Quality
No full textBackground: Code review is essential for ensuring software quality, but traditional review processes are time-consuming and may miss issues without human expertise. AI-driven tools such as Windsurf, Github Copilot, Claude Code, and Cursor have emerged to automate routine checks and improve productivity, yet empirical comparisons between different AI-driven tools remain limited. Objectives: This study evaluates how multiple AI-driven code review tools differ in improving developer productivity and software quality, and how developers perceive their ability to detect context-dependent issues compared to traditional non-AI review approaches. Methods: A controlled experiment was conducted comparing four AI-driven tools against a traditional non-AI automated baseline using identical code review tasks. Quantitative metrics included reviewtime, bug detection rate, and SonarQube-based quality indicators. Qualitative insights were gathered through semi-structured interviews to assess usability and context sensitivity. Data were analyzed using ANOVA, Tukey HSD, t-tests, and thematic analysis. Results: AI-driven tools significantly reduced review time (≈40-50% faster) and detected substantially more bugs than the traditional method. Improvements in code quality metrics - such as reduced complexity, fewer code smells, and lower security vulnerabilities - were consistent across all AI tools,with only minor differences between them. However, qualitative feedback revealed concerns about limited context-awareness and over-reliance on automated suggestions, indicating that AI may miss nuanced logical or architectural issues. Conclusions: AI-driven code review tools enhance efficiency and routine issue detection compared to traditional automated approaches, but they remain limited in handling context-dependent concerns. A hybrid strategy that combines AI efficiency with human judgment is recommended to achieve both high productivity and comprehensive software quality
Artificial Intelligence in Product Development : A Catalyst for Sustainable IT Practices for Business
No full textThe increasing demand for products and services coupled with growing environmental concerns has necessitated a shift towards sustainable product development. Traditional methods often prioritize functionality over environmental impact, leading to resource depletion and waste generation. To address this, as environmental concerns are increasing in importance, innovative solutions are required to integrate sustainability considerations into product lifecycles. This study investigates the role of Artificial Intelligence (AI) in promoting sustainability within product service systems. A systematic literature review was conducted to identify key AI technologies and methodologies employed across different stages of product development. The analysis focused on the impact of these technologies on environmental sustainability and business performance. The findings reveal that AI technologies, including machine learning, natural language processing, and virtual prototyping, can significantly enhance sustainability. These tools may optimize product design, reduce material consumption, and minimize environmental impact. Furthermore, AI applications in predictive maintenance, end-of-life management, and energy efficiency contribute to resource optimization and waste reduction. AI has the potential to transform product service system development by integrating sustainability principles. By optimizing resource utilization, reducing waste, and enhancing decision-making, AI can drive both environmental and economic benefits. While challenges such as data quality and algorithm development exist, the overall positive impact of AI on sustainability is evident.
AI-based fault localization approach for SCADA systems
No full textBackground: SCADA (Supervisory Control and Data Acquisition) systems are fundamental to the operation and stability of critical power infrastructure, such as electrical grids. However, fault localization in SCADA systems poses significant challenges due to their heterogeneous nature, comprising tightly integrated hardware and software components, and the sheer volume of data generated during their operation. The interplay between diverse system elements, including sensors, communication protocols, control algorithms, and monitoring software, adds layers of complexity to fault identification and resolution. Traditional fault localization methods—relying on manual log analysis, code reviews, and bug triaging—are often inefficient and struggle to scale with the increasing volume and complexity of SCADA environments. While Artificial Intelligence (AI)-based approaches have demonstrated potential in other domains, their application in the power industry, particularly in SCADA systems, remains underexplored. Objectives: This thesis aims to design, implement, and evaluate an AI fault localization approach tailored for SCADA systems, focusing on improving fault localization and reducing the number of bugs that propagate to production environments. The key innovation lies in guiding pre-trained AI models with domain-specific knowledge derived from SCADA-specific data sources, such as industry-specific bug reports, system logs, and work item histories. Methods: Employing the Design Science Research Process (DSRP), the research begins with problem identification through literature review and expert consultation to understand the limitations of traditional methods and identify opportunities for AI. In the solution design phase, pre-trained AI models are adapted to process SCADA-specific data using techniques such as Retrieval-Augmented Generation (RAG). By integrating historical and operational knowledge, the models are equipped to generate actionable insights tailored to the SCADA domain. The prototype is then empirically evaluated within a SCADA development environment, focusing on metrics such as accuracy, efficiency, and feedback from industry professionals. Results: The Power-RAG prototype, designed for fault localization in SCADA systems, was evaluated across two iterations. In the first iteration, open source PrivateGPT achieved 100% fault localization accuracy but was notably slow, averaging 88 seconds per query. To address this, a custom UI was developed, achieving an impressive 95% accuracy while significantly reducing query time to just 12 seconds—a stark contrast to the 343 seconds required by traditional manual methods. The prototype efficiently provided Area Path suggestions and actionable solution insights, that could lead to improved operational efficiency. Feedback from five industry professionals praised the user-friendliness, adaptability, and speed of the custom UI, while highlighting areas for improvement, including query sensitivity and robustness in handling diverse fault scenarios. These results underscore the balance achieved between speed and accuracy, making the Power-RAG a usable initial prototype for SCADA fault localization. Conclusions: This thesis explores the application of AI-driven fault localization methods within SCADA systems, an area where such implementations remain largely underexplored. By leveraging pre-trained AI models guided with SCADA-specific knowledge, this research demonstrates how these tools can effectively process complex datasets, such as system logs and bug reports, to identify and localize faults. The results show clear improvements in fault detection efficiency, accuracy, and overall system reliability compared to traditional manual approaches. While the findings highlight the feasibility and potential benefits of AI in enhancing fault localization workflows, it is important to acknowledge that this work represents an initial step rather than a comprehensive solution. The prototype developed in this thesis provides a foundation for further refinement and adaptation.
Advanced Failure Classification Models for Construction Machinery: A Case Study with Volvo CE : Integrating Machine Learning to Reduce Downtime and Operational Costs
No full textThis thesis looks at how artificial intelligence (AI) and machine learning (ML) can be used together to create better ways to classify failures in construction equipment, especially Volvo Construction Equipment (VCE). To keep machine downtime and operational costs as low as possible while dealing with the problems caused by class imbalances in the datasets, the goal is to switch from reactive to predictive strategies. Objectives The primary objective of this research is to design and evaluate sophisticated machine learning algorithms that analyze sensor data to classify potential failure types. The study also wants to find out how well deep learning models, like transformer-based architectures, work and how data balancing techniques can make failure classification systems more reliable. Methods: The research employs a quantitative analytical framework using real-world performance datasets from manufacturing equipment. We tested three different methods: (1) using unbalanced raw data as a starting point; (2) using SMOTE to balance the dataset, including a limited version for multiclass classification; and (3) using binary classification on the sampled data from Volvo CE. These methods enabled the exploration of the effects of data balancing on model performance and interpretability. Results The experiments highlighted the challenges posed by class imbalance and its adverse effects on the accuracy and reliability of the model. SMOTE significantly improved precision, recall, and F1 scores for underrepresented failure types. However, rare failure modes still present unresolved challenges. Transformer-based architectures demonstrated notable accuracy improvements, especially when combined with balanced datasets. Conclusions This study shows how important it is to fix class imbalances in failure classification datasets to make models more reliable and improve how well they work. The findings contribute to the advancement of AI-driven failure classification in the construction industry, paving the way for proactive maintenance strategies that reduce downtime and optimize costs
Exploring predictors of the five-time sit-to-stand test based on cross-sectional findings from the Swedish National Study on Aging and Care (SNAC)
No full textBackground As we age, staying physically active and reducing sedentary behavior becomes crucial. To understand how to achieve this, factors related to daily physical function such as five-time sit-to-stand (STS) time should be explored. This study aimed to investigate the associations between STS time, self-rated physical activity, physical function, health-related quality of life, physical and mental health in community-dwelling older adults aged >= 60 years. Method Cross-sectional design with self-reported and objectively measured data from adults aged >= 60 years (n = 819), acquired from the Swedish National Study on Aging and Care. Data was analyzed through multiple linear regression. Results The model (R-2 = 0.268) showed that STS time was predicted by grip strength (beta' = -0.204, p < 0.05), age (beta' = 0.202, p < 0.05), health-related quality of life (beta' = -0.192, p < 0.05), having fallen within the preceding twelve months (beta' = -0.127, p < 0.05), physical activities of perceived light to moderate intensity (beta' = -0.121, p < 0.05), one-leg stand (beta' = -0.099, p < 0.05), and education level (beta' = -0.092, p < 0.05). For STS time, health-related quality of life (beta = -0.354, confidence interval [CI] (-0.509)-(-0.199)), having fallen within the preceding twelve months (beta = -0.222, CI (-0.365)-(-0.078)), and physical activities of perceived light to moderate intensity (beta = -0.166, CI (-0.278)-(-0.053)) were the most prominent predictors. Conclusion The model highlights the importance of grip strength and health-related quality of life in predicting STS time in older adults. Clinicians can use these insights to develop interventions that maintain physical function by regularly assessing and monitoring these factors. Future research should explore the relationship between fall history, faster STS time, and the impact of grip strength and health-related quality of life on sedentary behavior among older adults.SNA
Exploring the Potential of Generative AI : Use Cases in Software Startups
No full textBackground and Related Work: Software startups face unique challenges in product development, including limited resources, the need for rapid innovation, and the constant pressure to adapt to market changes. Generative Artificial Intelligence (GenAI) has recently gained significant attention, offering capabilities to assist creative processes, generate content, and enhance decision-making through data analysis. However, how GenAI can be integrated into agile product development processes in software startups remains an open question. Objective: This study aims to identify potential use cases for GenAI in software startups and explore how GenAI can support innovation, overcome development challenges, and integrate with agile practices to improve product quality and development speed. Method: We identified a list of GenAI use cases from existing systematic literature reviews and mapped them to engineering process areas in software startups. Following that, we conducted workshops with experts to validate our results. Results: The results provide a descriptive overview of GenAI’s potential applications in software startup environments. Given the current state of the art, we identified areas that could benefit faster from integrating GenAI. Conclusions: The study delineates the prospective impact of GenAI on agile product development in software startups, showcasing areas of immediate applicability.