Publikationer från Stockholms universitet
Not a member yet
31237 research outputs found
Sort by
Mehrsprachige Ansätze in einemschwedischen DaF-Lehrwerk fürJahrgangsstufe 7
No full textA significant proportion of Swedish pupils originate from multilingual backgrounds; however,the education system in Swedish compulsory schools remains monolingual in its approach.The present paper examines the implementation of plurilingual and tertiary language didacticsin an educational material designed for German as a foreign language students in the 7thgrade. The article explores how to help students use all their languages while learning aforeign one.Ein beträchtlicher Teil der Schüler:innen in Schweden hat einen mehrsprachigenHintergrund, doch ist das Bildungssystem der Pflichtschulen nach wie vor einsprachigausgerichtet. Dieser Aufsatz wird die Umsetzung einer mehrsprachigen undtertiärensprachdidaktik in einem Unterrichtsmaterial für Deutsch als Fremdsprache in der 7.Klasse untersucht. Es wird analysiert, wie man den Schüler:innen den Einsatz aller Sprachenbeim Fremdsprachenerwerb erleichtern kann
Evaluating the Utility and Diversity of Synthetic Clinical Text Generated Using LLMs on MIMIC-IV Text
No full textElectronic Health Records (EHRs) are valuable for clinical research and AI applications, but privacy regulations such as HIPAA and GDPR restrict access to real data. Large Language Models (LLMs) offer a potential solution by generating synthetic clinical text. This study evaluates synthetic text produced by fine-tuned LLaMA3, Mistral, and Qwen models trained on MIMIC-IV clinical notes, focusing on utility and diversity. Specifically, the research question guiding this work is: “How does synthetic clinical text generated using LLaMA3, Mistral, and Qwen models that are fine-tuned on real clinical text compare in terms of Utility and Diversity to real clinical text (original MIMIC-IV text) when used as training data for clinical NER?” To address this, utility was assessed by training BERT-based Named Entity Recognition (NER) models on synthetic data and testing them on real records, while diversity was measured through vocabulary coverage, entity variety, and structural fidelity. Results show that models trained on text generated by LLaMA3 and Qwen achieved macro F1-score of 0.9451 and 0.9446 with accuracies 0.9788 and 0.9792, only slightly lower than real-data baselines, whereas Mistral performed less effectively. In terms of diversity, Qwen demonstrated richer vocabulary, more balanced entity distributions, and stronger long-tail representation, while LLaMA3 achieved the highest structural similarity but with repetitive patterns. Mistral showed limited lexical improvements but weaker structural alignment. Overall, the findings indicate that synthetic clinical text from LLaMA3 and Qwen provides strong utility and meaningful diversity, supporting its potential as a privacy-preserving substitute for real EHRs in clinical NLP tasks such as disease and drug entity recognition. Consistent with prior studies, showing that LLM-based synthetic data can enhance model performance while maintaining privacy, this work further advances the field by comparing multiple open-source models and revealing differences in utility and diversity
Rus och rang : Alkoholkultur och social position i Patrick Hamiltons Hangover Square
No full textUppsatsen undersöker alkoholens gestaltning och funktion i Patrick Hamiltons roman Hangover Square (1941) med utgångspunkt i ett ekonomikritiskt och sociologiskt perspektiv. Analysen fokuserar på hur alkoholbruket hos huvudpersonerna George Harvey Bone och Netta Longdon kan förstås som klassbetingade konsumtionspraktiker, och hur dessa i sin tur speglar olika former av habitus, social positionering och symboliskt kapital. Teoretiskt utgår studien från Pierre Bourdieus begrepp om fält, habitus och kapital, med stöd i Donald Broadys tolkningar, för att synliggöra hur smak, livsstil och kroppsligt inarbetade mönster samverkar i romanens skildring av social marginalisering och klassmässig distinktion. Genom en kvalitativ, kronologiskt strukturerad analys av fyra centrala scener visar uppsatsen hur alkohol fungerar både som ett medel för social tillhörighet och som en indikator på klassmässigt utanförskap
“It’s Never Not Political When It Comes to the Eurovision Song Contest” : A Reception Study on How Political Implications Affect European Viewership and the Vision of ‘Uniting by Music’
No full text This thesis examines the viewership of the Eurovision Song Contest in relation to its 2024 edition, which showcased a range of issues that impacted viewership. These issues included controversy about the exclusion of Russia, the inclusion of Israel amidst the Palestine conflict, and the disqualification of the Dutch artist Joost Klein. Using a mixed method research design of netnography of a Reddit online community and semi-structured interviews with Eurovision viewers, this thesis aims to understand how political presence in the ESC impacts viewer opinions, expectations, and attitudes. Key results show that both Redditors and interviewees tend to showcase a highly critical stance towards how the competition has developed. They express emotions that the competition has largely developed negatively and that this has altered their opinion and also their watching attitudes. A significant part of interviewees expresses that they were undecided and had lost interest due to various factors, indicating a concerning trend of declining appeal. These factors included that they felt left in the dark about the ESC’s political decisions on how they treated the participating countries, that the entertainment factor had declined, and that the competition was a representation of inequality and unfairness. It was recognised that for viewers, the ESC is much more than an international singing competition but a beacon that holds endless opportunities for European empowerment, community, and connection. Consequently, Europeans tend to express that the ESC has lost sight of these core values and call for redirection. Eventually, this thesis contributes to the body of knowledge about the ESC by contributing a contemporary perspective that reflects clearly how current-day viewership positions itself, contributing effectively to the larger debate on how the ESC would benefit from the implementation of viewers’ ideas, as it promotes itself as an event tailored to viewers
Birthfluencing : Influencing opinions on the practice of birth and homebirth on Instagram
No full textThis thesis investigates how Swedish midwives and doulas who are active on Instagram communicate about homebirth and childbirth, and whether their narratives align with official healthcare policies. The study uses thematic analysis of content from the Instagram accounts of these midwives and doulas to identify the themes that emerged. The findings indicate that certain central themes are consistent with official policies, while other themes critique Swedish birthing practices. Personal autonomy and risk perspectives from healthcare policies were mostly reflected in the themes. The results highlight tensions between the midwives, doulas, and the official medical authority. The study aims to contribute to understanding how social media can influence public perceptions of healthcare and showcase the differing values and critiques shared by social media influencers on this specific topic.
Sore losers on election night? Examining the temporal emergence of the winner–loser gap
No full textDespite numerous studies demonstrating a winner–loser gap in voter satisfaction with democracy, the temporal dynamics of its initial emergence remain poorly understood. This study introduces a new framework, distinguishing between short- and long-term effects. Short-term effects refer to immediate attitudinal shifts among winners and losers triggered directly by the establishment of the result. Meanwhile, long-term effects unfold gradually thereafter, through processes like cue-taking and motivated reasoning, polarizing voter attitudes over an extended period following the election. Leveraging survey data coincidentally collected during elections, the empirical analysis provides evidence for both effects. However, it is the long-term mechanisms that predominantly explain the gap’s emergence. Notably, in elections where a government transition occurred, a new gap did not materialize until a few weeks after the outcome solidified. These findings illustrate the temporal complexity of voter reactions, carrying significant implications for addressing challenges related to system support during election periods
Who got tested and who got sick? Sociodemographic inequalities in COVID-19 testing and hospitalization among 1.48 million individuals in Sweden
No full textIn the early stages of the COVID-19 pandemic, PCR testing served different purposes for individuals and for policy makers. Policy makers relied on testing for representative case numbers to track and mitigate the spread of the disease whereas individuals needed tests to protect themselves and others, or to travel or go work. Systematic differences in testing across population groups can bias case numbers, making it more difficult for policy makers to implement effective non-pharmaceutical interventions. We link records of 494 699 PCR-tests taken between 2020-07-01 and 2020-12-31 to individual records in several administrative registers for 1 480 126 working age individuals in the counties of Stockholm and Scania in Sweden. We estimate the likelihood of getting tested, test positivity rate and hospitalization risk by sex, household size, migration background, education, income and medical risk factors in the individual or in the household using regression models with age, occupation and neighbourhood as fixed effects. We find that testing behaviour vary independently by several demographic, socioeconomic and medical factors. Several groups that were at an elevated risk of being hospitalized for COVID-19, including men, individuals born outside Europe and those with low education, had low testing rates and high positivity rates. Numbers of confirmed SARS-CoV-2 infections reflect both infection rates and the testing behaviour of the population. To improve the utility of testing in future pandemics, policy makers may collect data on negative tests and dedicate part of the testing capacity for representative screening
KOR Från klimatbovar till klimathjältar? : Kan adaptivt fållbete bli en del i att lösa/mildra våra klimatutmaningar?
No full textGlobala mål, lokala vägar – Agenda 2030 i två kommunala verkligheter
No full textAgenda 2030 is a global framework with the goal of promoting sustainable devolopment through the implemenation of 17 global goals. These goals are estabalished at a global level and are then to be implemented at the local level, which gives municipalities a central role. The conditions for municipalities to integrate the sustainabilty goals in practice differ. The purpose of this study is to analyze how two municipalities with different conditions work towards the same global goals by answering how the implementation is carried out and which factors it is based on. A qualitative research approach has been used in which a comparative document analysis was conducted on Härjedalen and Österåker. Public documents have been reviewed through contet analysis, and the study is based on the theoretical framework of multilevel governance. The implemenation is influenced by geographical, economic and structural factors. The conclusion is that capacity, resources and geographical location determine how Agenda 2030 can be implemented at the local level. A global framework like Agenda 2030 does not automatically lead to change but requries cooperation between different levels, where local conditions must be considered in the implementation. The local level plays an important role in achieving the global goals, while its position involves limitations, challenges, and vulnerability in the development process. Keywords: Agenda 2030, sustainable devolopment, multilevel governance, local implementation, collaboration, rural, urba
Design and Implementation of an Integrated Data Processing and Visualization Center for Enhanced Traceability in Automated HIL Testing
No full textBackground Ensuring traceability between test results and requirements is a key challenge in the automotive industry as automated Hardware-inthe-Loop (HIL) testing becomes more widespread. At Volvo Car Corporation, automated HIL testing currently lacks an effective method to link test outcomes from AutomationDesk with requirements managed in CarWeaver. This gap creates inefficiencies in validating requirements and supporting compliance with ASPICE standards, particularly Base Practices 7 and 8 under the SWE.5 process group. Objective The primary research question addressed is: How can an integrated data processing and visualization center be designed to enhance traceability, consistency, and visibility in automated HIL testing and software integration processes? This study aims to develop a data processing and visualization center that automates traceability analysis, performs consistency checks, and provides visualization of requirement coverage across software releases. Method This research adopts a Design Science Research methodology following five sequential activities: explicating the problem, defining requirements, designing and developing the artefact, demonstrating the artefact, and evaluating its effectiveness. The approach combines case study strategies, stakeholder interviews, and prototyping within the Integration and Release team at Volvo Car Corporation. Implementation The developed artefact is an integrated data processing and visualization center that automates data extraction from AutomationDesk and CarWeaver, performs consistency checks between test cases and requirements, and generates structured outputs for visualization. The solution is fully integrated with Jenkins pipeline, enabling automated operation within existing workflows without requiring changes to existing tools. Result The artefact successfully addresses the identified traceability gaps by generating traceability tables in test reports, detecting inconsistencies between test cases and requirements, and providing Power BI dashboards that visualize requirement coverage and validation status across software releases. The solution demonstrates enhanced traceability, supports compliance with ASPICE Base Practices 7 and 8, and improves transparency in the software integration and release process. Conclusion This study demonstrates that a data-centric and tool-independent framework can effectively address traceability challenges in automated HIL testing. The solution reduces manual effort, improves consistency through automated checks, and offers modularity where each component can be run independently. Limitations include dependency on available APIs and the post-processing nature of the analysis. Future work may focus on extending the framework to other disconnected tools, improving long-term data retention, and applying the approach beyond the current team context