Publikationer från Högskolan i Jönköping
Not a member yet
26709 research outputs found
Sort by
From vision to everyday practice : A qualitative study on employee`s experiences of AI implementation withinJönköping municipality
No full textBackground: Artificial intelligence (AI) is transforming the labor market and influencing howorganizations structure work and make decisions. In the public sector, where change processes areoften complex and slow, it is crucial to understand how AI is implemented and experienced byemployees. This thesis explores how employees with HR-related responsibilities within JönköpingMunicipality perceive the implementation of AI in their daily work. Aim: The aim of this study is to describe how employees working with HR-related tasks at JönköpingMunicipality perceive the AI implementation process. Method: The study has mainly a deductive approach and is based on a qualitative case study involvingsix semi-structured interviews. The empirical data was analyzed using thematic analysis. Thetheoretical framework includes Self-Determination Theory (SDT), Lewin’s Change ManagementModel, and Kotter’s 8-Step Change Model. However, it has inductive elements as neither the codingprocess nor the third research question are based on previous theories. Conclusion: The findings show that AI implementation is currently at an early and unstructured stage,where individual initiative often replaces formalized processes. Three central themes were identified:(1) lack of structure in implementation, (2) positive attitudes and a supportive organizational culture,and (3) awareness of risks but limited knowledge. Although employees express a positive attitudetoward AI, the study concludes that clearer communication, targeted training, and a shared strategyare necessary for sustainable and inclusive AI integration. The study offers practical recommendationsfor the public sector and contributes theoretically to research on change management and the impactof technology on HR work
Adaptive Threshold Optimization for Radar-Based Human Presence : Optimizing Radar-Based Presence Detection for Real-World Applications
No full textThis project focuses on improving presence detection in smart buildings by utilizing radar-based sensors as an alternative to traditional PIR solutions. Passive infrared sensors often fail to detect stationary individuals and have limited coverage, making them unsuitable for reliable occupancy detection. To address these limitations, an Acconeer radar sensor is integrated into an embedded system platform. The system collects and processes raw radar data to identify human presence more accurately. A key component of the project is the development and evaluation of calibration algorithms that adapt to different environments and reduce false detections. These algorithms enable the system to dynamically adjust sensitivity based on environmental noise and usage patterns. Through iterative testing and data analysis, the aim is to create a robust, low-power solution for presence detection that can be applied in real-world smart building scenarios, contributing to both energy efficiency and user comfort. Thanks to Knowit AB and JTH (Jönköping Tekniska Högskola) for providing the hardware, and thanks to our supervisors Robin Davidsson, Fredrik Sjölander and Andreas Axelsson for feedback and guidance
Retrieval-Augmented Generation (RAG) för automatisering av jordbruksblockannoteringar : AI-driven bearbetning av strukturerad text och gränsdata för generering av beskrivande kommentarer
No full textManual annotation of agricultural block boundary modifications is a time-consuming task often prone to inconsistency and administrative inefficiency. This thesis investigates how retrieval-augmented generation (RAG) can be used to automate the generation of internal annotations (noteringar) based on structured spatial and administrative data from Jordbruksverket’s Basen system. The purpose was to evaluate whether AI-generated comments could align with human-written annotations in both tone and content, and whether such a system could reduce the burden of manual documentation in public administration workflows. A microservice architecture was developed, integrating PostgreSQL, FastAPI, and RAGnar—a retrieval system used to dynamically query historical block data. Structured prompts were constructed using fields such as status, granskorsak, area_difference, and direction_change, and passed into a locally hosted LLaMA 3 language model to generate descriptive annotations. Evaluation was per- formed using BERTScore to measure semantic similarity between the AI-generated and human written annotations across 1,974 matched pairs. Results showed an average BERTScore F1 of 0.6594, indicating moderate semantic overlap. A deeper qualitative comparison revealed that the system performed well in routine cases involving spatial changes and area updates, while struggling with annotations requiring legal reasoning, policy references, or context not captured in structured metadata. The findings suggest that while the method does not fully replace expert judgment, it can support case officers by generating draft annotations that follow institutional tone and format. This contributes to increased standardization, improved traceability, and potential time savings in annotation workflows. The approach may also be applicable in other domains that rely on structured spatial data and formaldocumentation, such as urban planning and environmental regulation.Manuell annotering av ändringar av jordbruksblockgränser är en tidskrävande uppgift som ofta är benägen att vara inkonsekvent och administrativt ineffektiv. Denna avhandling undersöker hur retrieval-augmented generation (RAG) kan användas för att automatisera genereringen av interna annoteringar baserade på strukturerad rumslig och administrativ data från Jordbruksverkets Basen-system. Syftet var att utvärdera om AI-genererade kommentarer kunde överensstämma med människoskrivna annoteringar både i ton och innehåll, och om ett sådant system kunde minska bördan av manuell dokumentation i offentliga förvaltningsarbetsflöden. En mikrotjänstarkitektur utvecklades, som integrerade PostgreSQL, FastAPI och RAGnar – ett hämtningssystem som används för att dynamiskt fråga historisk blockdata. Strukturerade prompter konstruerades med hjälp av fält som status, granskorsak, area_difference och direction_change, och skickades till en lokalt hostad LLaMA 3-språkmodell för att generera beskrivande annoteringar. Utvärderingen utfördes med BERTScore för att mäta semantisk likhet mellan de AI-genererade och människoskrivna annoteringarna över 1 974 matchade par. Resultaten visade en genomsnittlig BERTScore F1 på 0,6594, vilket indikerar måttlig semantisk överlappning. En djupare kvalitativ jämförelse visade att systemet presterade bra i rutinmässiga fall som involverade rumsliga förändringar och områdesuppdateringar, samtidigt som det kämpade med anteckningar som krävde juridiska resonemang, policyreferenser eller kontext som inte fångades i strukturerad metadata. Resultaten tyder på att även om metoden inte helt ersätter expertbedömningar, kan den stödja handläggare genom att generera utkast till anteckningar som följer institutionell ton och format. Detta bidrar till ökad standardisering, förbättrad spårbarhet och potentiella tidsbesparingar i anteckningsarbetsflöden. Metoden kan också vara tillämplig inom andra områden som förlitar sig på strukturerad rumslig data och formell dokumentation, såsom stadsplanering och miljöreglering
Rendering Matters: SSR vs CSR in Modern Web Development
No full textThis study investigates the critical differences between Server-Side Rendering (SSR) and Client-Side Rendering (CSR) in web development, examining both technical performance metrics and user perception. As websites become increasingly interactive and content-heavy, the rendering method decision significantly impacts user experience and engagement, particularly for the estimated 40% of global mobile users still operating on 3G or slower networks (GSMA, 2023). Through a comprehensive mixed-methods approach, two functionally identical web prototypes were developed and evaluated under optimal and throttled 3G network conditions. Quantitative performance analysis revealed that SSR outperformed CSR in initial loading metrics. However, CSR demonstrated superior performance in interaction responsiveness once fully loaded. The user study with 10 participants showed a strong preference (60%) for SSR under slow network conditions, with users citing "immediate content visibility" and "reduced perception of waiting" as key factors influencing satisfaction. This research bridges the gap between technical performance metrics and subjective user experience, demonstrating that perceived performance significantly impacts overall user satisfaction. The findings provide evidence-based guidance for web developers when selecting rendering strategies based on specific use cases and target audience network environments
A Comprehensive Evaluation of Cloud Storage Access in Flutter Applications – Direct and API Based
No full textEfficient access to cloud hosted data is an important factor for modern cross-platform applications. Developers working with Flutter must decide between different data retrieval strategies, these include direct database connections and API based intermediaries. However, existing studies often overlook key factors such as PaaS environments, network variability, cross-platform feasibility, and scalability under load. This thesis offers an evaluation of directly connecting to the Azure SQL database through the server, versus a RESTful ASP.NET Core API in a Flutter assessment app. Two application versions were created, one issuing SQL commands directly to the server using an MSSQL library, the other routing requests through an API; both were then subjected to tests simulating diverse network conditions and concurrent user activity. The investigation also examined platform compatibility for direct access and reviewed caching strategies from theliterature to understand their effects on performance and scalability. The findings highlight that secure direct connections from Flutter to the Azure SQL database are unreliable on the Web and Android versions of the app, effectively requiring an API layer to meet security and compatibility demands. However, direct access was successfully established on Windows, where all subsequent performance and scalability tests were conducted. Then a review of the literature examined caching strategies and their impact on performance and scalability. Results revealed that although direct access delivers faster responses in single-user, ideal network scenarios, API mediated interactions maintain consistent performance as user loadgrows and allow for centralized caching to reduce backend load. Both methods slow down inpoor networks, but the additional network hop of the API may contribute to a larger increase in absolute latency. Overall, while direct database connections can offer latency advantages in isolated cases, the API-based approach emerges as the more scalable and practical solution for Flutter apps interfacing with cloud databases. Its strengths in managing concurrent requests, conserving client resources, simplifying maintenance, and ensuring platform-agnostic security support a clearer path for developers aiming to build scalable applications
The Potential Application of Intelligent Automation to Replace Manual Tasks : A case study investigating task characteristics to evaluate the potential suitability of Intelligent Automation.
No full textPurpose: The purpose of this study was to investigate the potential of Intelligent Automation (IA) to replace tasks traditionally performed manually. This involved identifying key task characteristics and examining their influence on the feasibility of IA application. Methodology: An abductive approach was adopted, and to fulfill the purpose of the study, a single case study with an embedded design was used. Data were collected through interviews with six individuals experienced in Logistical Support Analysis (LSA) and three individuals with expertise in IA, in order to gain in-depth insights into both the nature of the tasks and the capabilities of IA. In addition, document studies were conducted on two separate occasions to enrich and validate findings. Key Findings: The findings identified six key characteristics, Repetitiveness, Resource Intensity, Human Error Susceptibility, Digital Nature, Data Volume and Data Quality, as central to determining a task’s suitability for IA. These characteristics align with the functional capabilities of Robotic Process Automation (RPA) and one or more Artificial Intelligence (AI) technologies which include Machine Learning (ML), Deep Learning (DL) and Natural Language Processing (NLP). Among these, Digital Nature was identified as a prerequisite, while Data Volume and Data Quality significantly influenced feasibility. Furthermore, combinations of the identified characteristics informed the selection of appropriate technologies, ranging from single data tools to integrated solutions combining RPA and AI. Contributions: This study extends the literature by identifying the characteristics Repetitiveness, Resource Intensity, Human Error Susceptibility, Digital Nature, Data Volume and Data Quality as suitable for IA. Additionally, the study provides a practical contribution by presenting a flowchart based on the findings, which serves as a decision-support tool for preliminary assessments of IA implementation potential. Limitations & Further Research: Due to the study’s single case design, the absence of cross- organizational comparison limits the generalizability of the findings. Therefore, further research could investigate whether the identified characteristics are equally relevant in other contexts
Exploring Methods to improve Context-Aware text Extraction and Categorization of a Large Language Model
No full textDebriefing training exercises and high-stakes operations often involves analyzing radiocommunicationto better understand interactions and decision-making. This research investigates methods forextracting and categorizing sender-recipient information from transcribed air traffic control (ATC) communication,aiming to support more effective analysis of such data. A quantitative experimental approachwas used to compare the accuracy of zero-shot and few-shot prompting techniques, as well as their finetunedcounterparts using Low-Rank Adaptation (LoRA).Zero-shot prompting provided a strong baseline for evaluating the effectiveness of few-shot andfine-tuned approaches. The few-shot method generated unexpectedly strong results, and its combinationwith a LoRA fine-tuned model achieved the highest accuracy. These improvements highlight how simpleadjustments can enhance a model’s ability to understand domain-specific communication.The findings suggest that these methods could benefit other high-stakes domains, such as militaryoperations, emergency response, and other uniformed services, by supporting faster and more accurateinterpretation of critical communications during both training and real-world events. The results aresupported by Fisher’s exact test, performed to determine statistical significance
The impact of greenwashing on customer trust and brand loyalty in FMCG marketing
No full textThis research focuses on greenwashing in the fast-moving consumer goods (FMCG) sector, researching whether greenwashing practices affect the trust of customers, their brand loyalty, and the way they make purchasing decisions. Greenwashing, the act of misleading consumers about the environmental benefits of products, has become a significant concern as the importance of sustainability efforts in marketing continue to rise. This study explores whether FMCG consumers notice sustainability claims, and whether they find the difference between genuine sustainability efforts and greenwashing important, or relevant to their purchase decision. The research builds on consumer behavior theories, significantly the theory of planned behavior (Ajzen, 1991) and signaling theory (Spence, 1973), to grasp how consumers process sustainability signals, such as eco-labels, and if these signals influence their trust and purchasing behavior. Using qualitative methods, the study uses focus groups to collect insights from consumers about their perceptions of greenwashing and if it affects their trust in FMCG brands. The findings suggest that while consumers might want to engage in sustainable behavior, their decision is very dependent on visual cues, social context and perceived effort. Trust in the claims is then not shaped by only the messaging of brands, but also by peer discussions and digital tools, which furtherly highlight how important credibility, accessibility and informal learning in FMCG marketing is when it comes to sustainability
