Publikationer från Uppsala Universitet
Not a member yet
127603 research outputs found
Sort by
Between Push and Pull Entrepreneurship : Understanding the Rise of Entrepreneurship Amid COVID-19 in the United States
No full text"What is the point of the Public Employment Service?" A discursive study of Facebook commentary on the Swedish Public Employment Service
No full textPrevious research has found that government agencies are increasingly adapting to the currentmedia logics, including embracing social media in order to connect with the public. Socialmedia enables organizations to express themselves through shorter, more digestible content andin more emotive, engaging manners. Less attention has been paid to how the interactiveaffordances of social media affect the interactions between public agencies and the public. Thisthesis explores how members of society interact with the Swedish Public Employment Servicethrough the comment sections of the agency’s Facebook profile. It delves into the researchquestions of how the comment sections function as spheres of discursive constructions, andwhat discourses that are created in these spheres regarding the trustworthiness of the agency.Comments from the comment section are used as data, and analyzed relying on the discursivemodel of Reisigls and Wodaks discourse-historic approach. The analysis is conducted throughthe lens of theories on deep mediatization, accountability, discourse construction and publicspheres. The findings suggest that comment sections are spaces where dispersed feedback andformations of alternative public spheres intertwine. In these spheres, a discourse is constructedthat calls into question the legitimacy and trustworthiness of the public agency. The resultsimply that comment sections are important spaces for marginalized communities to cometogether and form so-called “counterpublics”, but also where powerful organizations can becalled into question. This offers insights valuable to the studies of the use of social media in e-government, but also for public working to establish trust with the public through the media
Optimising a Soundness-First Numerical Analysis Framework : A Performance Study of Daisy
No full textDaisy is a framework for verifying numerical programs that prioritises soundness over runtime performance. This thesis investigates how much performance remains untapped due to this design choice during development. Through systematic profiling of Daisy's benchmarks, key bottlenecks were identified, including greatest common divisor calculations, hash code evaluations, and data structure operations. Targeted optimisations were implemented, including Stein's algorithm for greatest common divisor computation, lazy hash code caching, and improved hash map operations. The results show substantial performance improvements without compromising soundness, with up to 89% faster runtime on some benchmarks while maintaining identical numerical outputs and behaviour. These findings demonstrate that a lot of performance remains untapped, but also that soundness-first development and performance are not mutually exclusive. Verification tools like Daisy can be designed with soundness as the primary concern and later optimised for practical performance, making them viable for real-world applications requiring both reliability and efficiency
How do the dimensions of the TOE framework influence the challenges of implementing AI agents into ERP-systems? : A semi-systematic literature review
No full textThis study investigates the challenges influencing the implementation of Artificial Intelligence (AI) agents in Enterprise Resource Planning (ERP) systems, using the Technology-Organization-Environment (TOE) framework as an analytical lens. Given the recent rise in AI integration within enterprise software, exemplified by platforms such as SAP Joule and Microsoft Copilot, understanding the enablers and barriers to successful adoption is essential. Through a semi-systematic literature review and thematic analysis of twelve academic sources, the study identifies recurring themes across the technological, organizational, and environmental dimensions of TOE. The study gives practical insights regarding the challenges when integrating AI into an ERP system. With comparisons to previous research, where the TOE framework has been used to explain technological adoption and assimilation of ERP systems, the challenges were found to explain the how rather than the why of technological integration. The study contributes to the theoretical development of the TOE framework in AI-ERP contexts and offers practical insights for stakeholders planning to integrate AI agents into ERP systems. Findings reveal that the technological context, particularly data quality, infrastructure readiness, and compatibility, plays a dominant role, often determining the possibility of AI integration. Organizational factors such as management structure, integration strategies, and employee management serve as supportive infrastructure, while environmental factors, including legal regulations and vendor dependency, act as conditional constraints.Denna studie undersöker de utmaningar som påverkar implementeringen av artificiell intelligens(AI)-agenter i ERP-system (Enterprise Resource Planning), med hjälp av ramverket Technology-Organization-Environment (TOE) som analytiskt verktyg. Med tanke på den senaste tidens tillväxt av AI-integrering inom affärssystem, vilket exemplifieras av plattformar som SAP Joule och Microsoft Copilot, är det viktigt att förstå möjliggörarna och hindren för att framgångsrikt kunna integrera dem. Genom en semi-systematisk litteraturundersökning och en tematisk analys av tolv akademiska källor identifierar studien återkommande teman inom de tekniska, organisatoriska och miljömässiga dimensionerna av TOE. Studien ger praktiska insikter om utmaningarna vid integrering av AI i ett ERP-system. I jämförelse med tidigare forskning, där TOE-ramverket har använts för att förklara teknologiskt införande och assimilering av ERP-system, fann man att utmaningarna förklarar hur snarare än varför teknologisk integrering genomförs. Studien bidrar till den teoretiska utvecklingen av TOE-ramverket i AI-ERP-sammanhang och erbjuder praktiska insikter för intressenter som planerar att integrera AI-agenter i ERP-system. Resultaten visar att den tekniska dimensionen, särskilt datakvalitet, infrastrukturens beredskap och kompatibilitet, spelar en dominerande roll och ofta avgör möjligheten till AI-integrering. Organisatoriska faktorer som ledningsstruktur, integrationsstrategier och hantering av medarbetare fungerar som stödjande infrastruktur, medan miljöfaktorer, som inkluderar lagar och leverantörsberoende, fungerar som villkorliga begränsningar
Child Labour During the Century of Change : A Study of Child Labour’s Evolution on the Countryside and in the Cities, Within a Few Swedish Industries
No full textThis paper analyses the growth and decline of child labour, and the causes of these changes, in the Swedish factories during the period 1830 to 1890. By studying the textile and paper factories in the cities and on the countryside through the records of the Kommerskollegium’s annual reports on the nation’s factories, this study has shown that the evolution of the size and basis for change of child labour were more nuanced and complex than previously shown. Not only between different industries and periods of time, but also whether the factory was in a city or on the countryside. For much of the 1800s, child labour was on the rise in the Swedish factories regardless of industry or location. Towards the end of the century however, child labour slowly faded away. Generally, the transition from growing to shrinking child labour was believed to have occurred during or around the 1850s, but this study has shown that not to be the case in every industry branch. It varied between the early 1840s to the late 1880s, in some sectors, with repeated growths and declines rather than a single cycle. The frequently accepted explanation for the evolution of child labour in the factories during this period has been that children’s profitability caused the expansion and regression of the use of children as labourers. Mechanisation, during its development, made children profitable because they allowed factories to better compete with other industries and when mechanisation was fully implemented it made children unprofitable, as their low wages no longer made up for the potential profits of employing an adult worker. This paper has shown this to be only a part of the truth, and varyingly so. Legislation against child labour, and poverty, also explained certain aspects of the evolution of child labour, and together with mechanisation gives a more complete picture of what determined the rise and fall of children in the factories. The process was not uniform, however, and different parts of the industries experienced different trends at different times. In conclusion, this paper attempts to study and explain the evolution of child labour in the Swedish factories. It shows that the story of child labour during the century of change was not one single story, but a plethora of different routes and causes. The study highlights how multiple theories can help explain different parts of the larger puzzle
Intelligent Data-Driven Chatbot for Enhanced Information Retrieval and Interaction
No full textThis thesis presents the design and implementation of an intelligent, data-driven chatbot aimed atenhancing information retrieval and diagnostic support within the automotive service domain. Ad-dressing the increasing complexity of modern vehicles and the limitations of traditional diagnosticmethods, the proposed system integrates Retrieval-Augmented Generation (RAG) with LargeLanguage Model (LLM) to create a context-aware, multilingual chatbot. Leveraging structuredand unstructured repair data from Volvo’s workshop records, the chatbot utilizes advanced datapreprocessing, semantic vector search via Azure AI Search, and prompt engineering strategiessuch as Chain-of-Thought reasoning. The system delivers real-time, grounded, and coherentresponses by dynamically retrieving relevant documentation and contextual data. Evaluated us-ing LLM-based metrics for context precision, faithfulness, and answer relevance, the solutiondemonstrates substantial improvements in diagnostic efficiency and user experience. This workcontributes to the growing field of Artificial Intelligence (AI)-powered maintenance tools, highlight-ing the potential of domain-specific conversational agents in complex technical environments
Energieffektivisering av kraftelektroniska system : Potentialen i spillvärmeåtervinning och lokal elproduktion hos STATCOM- och HVDC-stationer
No full textSyftet med rapporten var att undersöka energieffektivisering av de kraftelektroniska systemen, HVDC (High Voltage Direct Current) och STATCOM (Static Synchronous Compensator). Rapporten undersökte potentiella användningsområden för spillvärme som genereras under stationernas drift, samt om den elkonsumtion som krävs för drift kan produceras inom stationernas yta. Möjliga tillämpningar för spillvärmen som undersöktes var fjärrvärme, elproduktion via organisk Rankine-cykel (ORC), uppvärmning av närliggande växthus och integrering i stationernas befintliga värme- och ventilationssystem (HVAC). Många av de undersökta värmetillämpningarna är tekniskt möjliga att genomföra, givet att stationerna är belägna i närheten av där spillvärmen ska utnyttjas så att konvektionsförluster vid värmetransport inte blir för stora. Gällande egenproduktionen av el undersöktes potentialen av att installera solceller och småskaliga vindkraftverk på stationernas tak. Resultaten indikerar att varken STATCOM- eller HVDC-stationer kan försörjas helt av lokal elproduktion med solceller och vindkraftverk, då deras egenkonsumtion är högre än den potential som finns för produktion bortsett från några få timmar under dagarna med högst elproduktion. Däremot kan den lokala elproduktionen användas i stationerna i stället för el från nätet, vilket avlastar nätet. Gällande spillvärmeåtervinningen tydde resultaten på att fjärrvärme kunde förstärkas till en viss mån. Växthus i anslutning till STATCOM- och HVDC-stationerna kunde dra relativt stor nytta av spillvärmen. Även organiska Rankine-cykeln kunde producera el, och HVAC-systemen kunde även stödjas.The purpose of the report was to investigate possible energy efficiency improvements of the power electronic systems, HVDC (High Voltage Direct Current) and STATCOM (Static Synchronous Compensator). The report investigated potential areas of use for waste heat generated during the operation of the stations, and if the electricity consumption required for operation can be produced within the stations' premises. The potential applications of waste heat that were explored included district heating, electricity production with organic Rankine cycle (ORC), heating of nearby greenhouses, and integration into the stations' existing heating and air conditioning (HVAC) systems. Many of the investigated waste heat applications are technically feasible, provided that the stations are located close to where the waste heat is to be utilized so that convection losses during heat transport are not too great. Regarding the electricity production, the potential of installing solar panels and smallscale wind turbines on the stations' roofs was investigated. The results indicate that neither the STATCOM nor the HVDC stations can be completely self-sufficient in locally produced electricity, since their usage is greater than the production, except the days with highest electricity production. However, the locally produced electricity can be used by the stations instead of the grid, partly reducing the load on the grid. Regarding heat waste recycling the results indicate that district heating can be partly strengthened. Greenhouses nearby the STATCOM and HVDC stations could be heated with the waste heat. Organic Rankine cycle was another waste heat solution which could produce electricity, and the HVAC systems could be supported as well
Exploring the Relationship Between Gene Expression and the Three-Dimensional Locus Positioning : A Microfluidics-based Study on the Escherichia coli Chromosome
No full textUnderstanding how chromosome structure influences function is a central question across alldomains of life. In eukaryotes, heterochromatin, which is tightly packed and transcriptionally inactive, differs from more often transcribed euchromatin. However, such a division has notbeen observed in prokaryotic chromosomes. The three-dimensional organization of prokaryotic chromosomes and its functional implications remain poorly understood, despite the relative simplicity of bacterial genomes. In this project, we aimed to analyze the relationship betweengene expression and the three-dimensional locus position over the cell cycle in the Escherichia coli chromosome through two approaches; (1) inducing high expression at several different loci across the chromosome to analyze the three-dimensional locus position pre- and post induction, and (2) investigating the locus-dependent expression by introducing a gene expression reporterat loci with different distributions. Tracking of chromosomal loci was performed using a microfluidic system and fluorescence microscope, to get single-cell resolution. The results demonstrated that genomic position where gene expression was induced affected both the spatial organization of the chromosomal loci and the physiological response of the cells. Furthermore, these effects varied between different loci. Additionally, introducing a gene expression reporter at different loci showed that expression levels vary depending on the position of the locus, implying locus-dependent expression. The findings of this project demonstrated that gene expression in E. coli is influenced not only by regulatory elements suchas promoters and transcription factors, but also by the genomic- and spatial position of the gene. These findings could improve our ability to engineer bacteria for biotechnological applications or improve the development of antimicrobial strategies that disrupt gene expression by targeting chromosomal organization
Siri says you're hired : En kvalitativ studie om förändring vid AI-implementering i rekryteringsföretag
No full textAI är något som blir mer och mer vanligt att implementera inom rekryteringsprocessen. Studiens syfte är att identifiera och analysera vilka drivkrafter som ligger bakom rekryteringsföretags beslut att implementera AI och att undersöka hur förändringsprocessen har sett ut inom de utvalda företagen, samt om processens utformning har haft en påverkan på eventuellt förändringsmotstånd. Våra frågeställningar är; “Varför väljer rekryteringsföretag att implementera AI?” samt “Hur har förändringsprocessens utformning påverkat eventuellt förändringsmotstånd inom rekryteringsföretag?”. Med teoretisk förankring i Lewins teori om förändring i tre steg har vi genomfört och analyserat intervjuer med anställda på två rekryteringsföretag. Resultatet avslöjar två primära drivkrafter för AI-implementering; effektivitet och behovet av konkurrenskraft. Vidare har möjligheten att avstå AI-användning minskat förändringsmotståndet men det kan komma att ändras när användningen blir obligatorisk. Detta är viktigt då många företag står inför arbetet med att implementera AI i verksamheten. AI is becoming more common to implement in the recruitment process. The purpose of this study is to identify and analyze which driving forces are behind recruitment agencies' decision to implement AI and to explore how change has transpired within the chosen companies, and whether the structure of the process has had an impact on possible resistance to change. Our research questions are; “Why do recruitment agencies choose to implement AI?” and “How has the structure of the change process affected possible resistance to change within recruitment agencies?”. Theoretically anchored in Lewin’s theory of change as three steps, interviews with employees of two recruitment agencies have been held. The results show two primary driving forces for AI-implementing; efficiency and competitiveness. The possibility to refrain from AI-use has reduced resistance to change, however this might change as the use becomes mandatory. This is of importance since many companies are facing the undertaking of implementing AI