Publikationer från Uppsala Universitet
Not a member yet
127603 research outputs found
Sort by
BEYOND BORDERS: FORECASTING SWEDEN’S GLOBAL TRAVEL BEHAVIOR IN 2045 WITH BAYESIAN STATE-SPACE MODELS
No full textInternational travel behavior is evolving rapidly, necessitating accurate forecasting models aiding infrastructure and policy planning. This study presents a Bayesian hierarchical model that jointly predicts the choice of international destination country and travel mode for Swedish travelers using Swedish National Travel Survey (RVU) data. The model integrates demographic, socioeconomic, and cost-related predictors, combining a multinomial logistic regression for mode choice with a conditional categorical model for destination. Key model features involve Bayesian imputations of missing household income data, random walk priors on mode intercepts, and a mode-specific Dirichlet parameter to compensate for overdispersion in the destination model. Posterior inference was performed using Hamiltonian Monte Carlo sampling. The model was validated with out-of-sample test data and additional external datasets to evaluate predictive performance. All out-of-sample observed values for mode choice fell within the model’s prediction intervals, indicating a well-calibrated model. Forecasts for 2045 suggest uncertainty in future travel patterns, particularly for car and air travel.
Artificiell Intelligens inom HR - användning, attityder & upplevd nytta
No full textSyftet med denna studie är att undersöka i vilken omfattning som Artificiell Intelligens (AI) används inom HR och på vilket sätt, samt vilka eventuella samband som finns mellan användning och upplevd nytta och attityder och upplevd nytta. Undersökningen grundas på en kvantitativ tvärsnittsstudie där data samlats in genom en enkät. Enkäten delades via sociala medier och fick sammanlagt 68 svar. Mätverktyget är inspirerat av Technology Acceptance Model (TAM). Data analyserades med hjälp av korrelationsanalys och ANOVA (variansanalys) med tillhörande post-hoc-test. Studiens resultat tyder på att AI används i stor utsträckning inom HR-sektorn, där 91% av respondenterna uppgav att de använder AI i sin yrkesroll. Studien tyder också på att HR-anställda som arbetar på operativ nivå använder AI oftare i sin yrkesroll, till skillnad från de som arbetar på strategisk nivå, och då främst till administrativa arbetsuppgifter och för att komma på lösningar till arbetsrelaterade problem. Studiens resultat tyder även på ett statistiskt signifikant samband både gällande HR-anställdas upplevda nytta av AI och deras faktiska användning samt ett samband mellan upplevd nytta och deras attityd till AI. Detta då de som använde AI mer frekvent tenderade att i högre grad instämma i frågorna kring att AI hjälper dem i arbetet samtidigt som en hög upplevd nytta av AI-baserade verktyg sammanföll med positiva attityder. Studien konstaterar att AI utgör ett väl använt arbetsverktyg inom HR och att acceptans och användning av AI har ett samband med båda indexen, upplevd nytta och attityd, vilket går i linje med Technology Acceptance Model (TAM). Studiens resultat indikerar på att AI är ett allt viktigare arbetsverktyg eftersom det används trots att det inte alltid är etablerat på arbetsplatsen. Vidare framkom att, trots den positiva inställningen till AI, inte lika många respondenter kände sig helt bekväma med tekniken eller upplevde den som lättanvänd, vilket pekar på behovet av fortsatt stöd och utbildning för att främja en trygg användning av AI inom HR-branschen
Kampen om inflytande vid rekryteringar i svensk elitfotboll : En kvalitativ studie om dataanalysens roll och kommunikativa legitimering i spelarrekrytering i svensk elitfotboll
No full textThis qualitative interview study examines the Swedish elite men’s football clubs AIK, IK Sirius FK, Västerås SK and Örebro SK with the aim of analyzing which actors hold the most power in player recruitment, and how data analysis is communicated throughout this process. The central findings highlight how the structure of the recruitment process and the legitimization of decisions are shaped through an interplay between human and non-human actors. Communication between these actors constitutes both the recruitment process itself and the use of data analysis within it. The study is based on semi-structured interviews, participant observation, thematic analysis of empirical material, and a theoretical framework combining Actor-Network Theory and the Montreal School’s Communication as Constitutive of Organization-perspective. This is complemented by previous research in the field. The theoretical foundation treats human and non-human actors as equally significant, and views reality as socially constructed through communicative processes. The player recruitment process is built upon an interaction between what this study refers to as subjective and objective scouting. Subjective scouting involves human observation through video or live match attendance, where gut feeling, intuition, and the ”football eye” are decisive. Objective scouting, on the other hand, centers on data analysis based on club-selected KPIs. To legitimize decisions, stability within the decision-making network is required, with core actors being club identity, financial resources, the sporting director, and a governing document referred to in the study as the ”player profiles”. These profiles constitute the most stabilizing actors in the network. Data analysis is communicated through several actors. The sporting director, scouts, and analysts all give ”voice” to data during decision-making. Moreover, through documents like the player profiles, data analysis gains significant material influence, as these profiles serve as reference points in many communicative exchanges. The profiles themselves are based on KPIs shaped by data, club identity, financial conditions, and the subjective assessments of human actors. Although data analysis is often presented as objective, in practice it is shaped by subjective choices, interpretations, and communicative translations. The study problematizes the notion of objectivity and demonstrates how it is constructed through interactions between various actors
Forcing the Hand : A Study of Stress and Persistent Narrative in Horror Game Design
No full textThis paper aims to explore how to craft what we term a persistent narrative in intentionally stressful gameplay situations. Persistent narrative in this case equals a memorable and understandable narrative by the player. We chose this field of study because it is relevant to our own experiences as game designers, and we hope the results will aid us and other game designers in future endeavours. The study was done through a prototype gambling game featuring bodily mutilation as a consequence for losing, and the narrative was communicated through environmental design and simple dialogue. With this research, we aim to understand how one can communicate a persistent narrative to the players, even while putting them under stress and delivering an intense gaming experience. The research was conducted through playtests, creative practice, and interviews, and our findings showed that stress is evoked in players best if they are aware and familiarised with the game's stress-inducing mechanics. Narrative-wise, it was concluded that even though a potential method of implementing narrative was found, the data was insufficient to make a finalised verdict on whether or not its effects are guaranteed. The reason for insufficient data is the fact that for our final playtest, we only had one tester.
: Stylistic value of the adjective and its position in the Spanish noun phrase
No full textSe dice comúnmente que la lengua española de un nivel estilístico más alto contiene más adjetivos antepuestos que la lengua menos estilística. La posición del adjetivo en el sintagma nominal ha sido rigurosamente investigada y existe una gran cantidad de teorías que explica por qué se coloca antes o después del sustantivo. Un tipo de adjetivos se pone tanto en la anteposición como en la posposición y, en cualquier caso, mantiene su significado. El propósito de esta tesina es investigar si se presenta correlación entre tales adjetivos de un nivel estilístico más alta y la anteposición, independientemente del nivel estilístico del texto. En esta investigación se compara la frecuencia relativa de ante- y posposición de parejas de sinónimos separados por el nivel estilístico en un análisis cuantitativo y luego se realiza un análisis cualitativo, con el fin de eliminar factores ajenos que podrían influir en el resultado. El resultado de la investigación muestra correlación entre alto valor estilístico del adjetivo y la anteposición y, por lo tanto, que el nivel estilístico del adjetivo tiene influencia en su colocación en el sintagma nominal español
Analyzing the Downstream Implications of Pretrained Image Representations in Visual Topic Models
No full textThe explosive growth of online content, particularly images, presents a significant challenge for analysis, as traditional text-based methods often fail to capture the full context and nuance of multimodal communication. This thesis addresses the necessity of effectively understanding vast image datasets by exploring the application of visual topic models. It investigates how different image representation techniques influence the quality and interpretability of automatically derived topics. The problem was tackled by developing a robust pipeline for visual topic modeling, employing three distinct image embedding methods: DINOv2 (purely visual), SigLIP2 (multimodal), and LLM-based embeddings (textual descriptions). A comprehensive evaluation was conducted, combining quantitative metrics for hyperparameter optimization with two novel human-centric crowd-sourced tasks designed to assess topic coherence and interpretability from both semantic and visual perspectives. Results indicate that the choice of image representation significantly impacts topic quality. DINOv2 consistently generated the most coherent and interpretable topics, demonstrating surprising strength in semantic understanding despite its purely visual training. SigLIP2 achieved a balanced performance, effectively capturing both visual and semantic relationships. Conversely, while LLM-based embeddings excelled semantically, their visual coherence was found to be lower. This work demonstrates the critical need for selecting appropriate image representation models based on the specific analytical goals. The findings provide valuable guidance for researchers and practitioners engaged in large-scale visual content analysis, enabling more accurate and insightful interpretations of complex online visual data
Development and psychometric evaluation of the 12-item Polish version of the Pain Vigilance and Awareness Questionnaire
No full textThe Pain Vigilance and Awareness Questionnaire (PVAQ) is widely used to assess attention to pain, but no validated Polish version has been previously available. This study aimed to translate the PVAQ into Polish and evaluate its psychometric properties. Two studies were conducted: Study 1 included 418 chronic pain patients recruited online who completed the scale alongside measures of depression, anxiety, pain catastrophizing, pain anxiety, pain intensity, functional limitations, and psychological flexibility. A follow-up was completed by 79% after 30 days. Study 2 involved 148 hospital-recruited chronic pain patients who completed the adapted scale and measures of pain intensity and functional limitations. None of the previously proposed models describing the factor structure of the PVAQ scale adequately fit the data. However, factor analyses indicated that a 12-item version with two correlated factors-Passive Awareness and Active Vigilance-provided acceptable model fit. The Polish PVAQ showed good internal consistency, moderate test-retest reliability, and expected correlations with related constructs. Interestingly, Passive Awareness (but not Active Vigilance) was positively associated with components of psychological flexibility. The Polish 12-item PVAQ is reliable and valid instrument for assessing pain-related attention in chronic pain samples, aiding healthcare professionals in identifying highly vigilant individuals to tailor interventions accordingly
Entanglement in the Quantum Ising Chain
No full textThis thesis studies entanglement in the transverse Ising chain. The Ising chain contains two main components: spin-spin couplings between particle spins and spin-field couplings between particle spins and an external magnetic field. As the relative influence of the two coupling-terms is varied with the field strength, a quantum phase transition between a magnetically ordered phase and a quantum paramagnetic phase occurs. Entanglement between nearest neighbor and next-nearest neighbor sites in the chain is quantified by concurrence and negativity at varying field strengths. These calculations were done numerically using Python. To compute this, the exact spectrum, correlation functions, magnetization, and the density operator are derived. The entanglement was found to be zero in the two phases of the system, while it had a maximum at or close to the phase transition. Criticality seemed to be closely connected to a singularity in the derivative of the entanglement with respect to the field strength in all cases except the next-nearest neighbor concurrence, where criticality seemed to be indicated by a maximum of the entanglement itself. Concurrence and negativity did not quantify entanglement equally, in particular they did not rank the ground states of the model at different field strengths by the amount of entanglement in the same order
On the Origins of Gaia-BH3: 3D Non-LTE Reanalysis of its Stellar Companion
No full textIn a paper by Balbinot et al. it is proposed that the Gaia BH3 binary system originates from the ED-2 disrupted stellar stream. While they provided strong kinematic evidence to support the claim, the chemical evidence displayed weaknesses. The aim of this project was to analyze the stellar ED-2 and companion star data used by Balbinot et al. in 1D LTE and 3D NLTE in order to confirm or deny the proposed origins. The data was analyzed through 1D LTE and 3D NLTE line-by-line scatter abundance plots and chemical abundance ratio plots. The abundance ratios of the stars were also compared to atomic diffusion models in order to determine whether potential abundance differences could be explained by atomic diffusion processes. The results showed that the abundance differences between the Gaia BH3 companion star and the ED-2 stars is significant even when analyzed in 3D NLTE. It was also determined that the abundance discrepancies are reduced but remain significant after considering atomic diffusion models from the literature. Stellar black hole formation theory presents us with scenarios that could potentially explain the origins of Gaia BH3, but we would need to know how the black hole originally formed. In conclusion, it was determined that according to the chemical compositions, it is unclear if Gaia BH3 in fact originates from the ED-2 stellar stream.I en rapport från Balbinot et al. påstods det att det binära Gaia BH3 systemet härstammar från den störda ED-2 stjärnhopen. De presenterade starkt kinematiskt bevis för detta, men det kemiska beviset var bristande. Målet med detta projekt var att analysera datan på ED-2 och följeslagarstjärnan som användes av Balbinot et al. genom 1D LTE och 3D NLTE modeller för att bekräfta eller motbevisa det föreslagna ursprunget. Datan analyserades genom 1D LTE och 3D NLTE linje-för-linje spridnings plottar och kemiska abundansförhållande plottar. Abundansförhållandena för stjärnorna jämfördes även med atomära diffusionsmodeller för att undersöka om eventuella skillnader kan förklaras av atomär diffusion. Resultaten visade att skillnaderna i metallicitet mellan följeslagarstjärnan och ED-2 stjärnorna är stor även genom 3D NLTE analys. Det bedömdes även att atomär diffusion inte kan förklara den stora skillnaden i metallicitet. Teorin för formationen av svarta hål ger oss scenarion som potentiellt kan förklara ursprunget av Gaia BH3, men för att kunna dra några slutsatser behöver vi veta mer om hur det svarta hålet bildades. Sammanfattningsvis bestämdes det att det, enligt de kemiska sammansättningarna, är oklart om Gaia BH3 härstammar från ED-2 stjärnströmmen
