No description available.

The purposes of this qualitative descriptive study were to describe the perceptions of rural, older Appalachian women who live alone regarding systems of informal and formal care and to understand if traditional cultural norms influence attitudes and decisions to access these two systems. Older Appalachian women in this study defined themselves and their health in terms of their homes and as women who care for themselves informally and value independence and privacy. Five major themes emerged from the data for informal care, and three related to formal care or use of it. The findings support a reconceptualization of informal and formal care and point out reasons why these women chose to use or not use these two systems of care. Furthermore, they reveal how changes in the formal care system could support health promotion and prevention strategies grounded in everyday ways of maintaining health within the context of home.

In the field of public diplomacy, it has increasingly become relevant to develop analytical operations in order to gain knowledge on the perceptions of foreign publics and their attitudes towards countries – a construct known as the country image. In recent time, research on public diplomacy has been increasingly occupied with the impact of media on country images due to the continuous expansion and fragmentation of the hybrid media landscape. Academics and practitioners alike must navigate through large quantities of data and different choices regarding prioritization of sources and methods in order to find suitable analytical frameworks to properly investigate the country image as an analytical object. This thesis addresses these analytical challenges by developing a diachronic text mining analysis of the Swedish country image in the British newspaper the Guardian. Sweden has in recent years drawn attention from the international media, for example during the so-called refugee crisis of 2015 and during the covid-19 pandemic of 2020. In the extensive media coverage of these major events the Swedish course of action was met with a wide range approval and criticism. Using a mixed-method approach of distant and close reading, this thesis approaches the news coverage of Sweden in the Guardian through a content analysis designed in three analytical steps: topic modelling, collocation and concordance analysis, and diachronic corpus assisted discourse analysis. In each of these steps, the appearance of different dimensions of the country image was explored using a dimensional model for integrative country image analysis developed by Ingenhoff and Buhmann. The design of the mixed-method approach showcase how large quantities of textual data can be analyzed in a new diachronic approach, bringing strains of research from digital humanities to the field of public diplomacy.

Internet search has been marked by the dominant use of query search, specifically Google, since the mid-1990s. The public release of the AI-based search tool, chatGPT, powered by a recent innovation in deep learning known as large language models (LLMs), marks a paradigm shift in internet search technology. While the essence of both the search technologies, namely, the retrieval of information from the internet, remains the same, there appears to be a marked difference in the manner of their use and perception by both the general public as well as in media. Prior studies have highlighted the importance of assessing perceptions of new technology on users. Examining the impact of this recently-introduced form of search compared to the original query-search can provide valuable insights into users’ perception of search technologies as well as identify underlying attitudes towards AI. This study investigates the distinct discursive patterns characterising user perceptions of these two search paradigms. It uses a collected text corpus of media articles and forum data as its research material, and employs Latent Dirichlet Allocation (LDA) topic modelling to generate a quantitative set of topics. These are then examined qualitatively through the lenses of technological frames and discourse analysis to uncover user perceptions. Findings indicate that user discourse patterns diverge, anticipatory themes differ and there is variation in user concerns as well as media coverage. This research contributes insights into evolving technological perceptions, societal consequences, and the media’s role in shaping user discourse. It also highlights that further investigations into the anthropomorphic aspects of digitalisation and the evolving information landscape may offer promising avenues for future research.

Combating crime is a complex task with cultural, political, and legal dimensions. In technologically advanced societies, surveillance technology can be used to aid law enforcement. A few examples of such tools are drones, cameras, and wiretaps to mention a few. As such tools become more commonplace, the need to address associated issues increase which relate to cultural, political, and legal dimensions and different stakeholders. Hence, the purpose of this thesis is to discern the impact of informatics research on surveillance technology and map out similarities and discrepancies between views of social media users, researchers, and professionals within law enforcement. The thesis impose a heuristic perspective and stem from both positivist and interpretivist tradition. The Panopticon metaphor and Panopticism are used as a theoretical lens, mainly to discuss and contextualize the findings. Data was from Twitter and Scopus by using scripts and by conducting an interview with law enforcement staff in Sweden. A total of 88 989 tweets and 4 874 research papers were retrieved. These were analyzed using topic modeling which assigned a dominant topic to each tweet and research paper. The interview was thematized using both the literature review and the topic modeling findings for guiding framework. The findings showed that there were seven topics found within the Scopus dataset and four topics within the Twitter dataset. It was found that privacy was one of the least mentioned aspects in all three datasets and that law enforcement personnel see it as closely related with efficiency. Military applications and usage were found in both research papers and tweets and law enforcement staff use a variety of ICT in their daily work. Based on the findings, it seems as though surveillance technology today can suitably be characterized as being bi-directional, both in the form of sousveillance and surveillance which relates to the Deleuzian perspectives on Panopticon. It was concluded that concrete implementations of surveillance technology attracted the most attention compared to more abstract themes such as ethics and privacy. But in all both datasets, specific ICT was addressed from a critical perspective. Similarly, law enforcement personnel viewed privacy and integrity from the organization’s perspective and highlighted rules and regulation. For future work, sentiment analysis is suggested to supplement topic modeling as well as imposing a longitudinal approach or adding additional social media sources.

This thesis embarks on a journey to investigate the discourse contained within the policy documents examined by utilizing the topic modeling technique labeled Latent Dirichlet Allocation. The aforementioned investigation will be conducted through the theoretical lens of Systems Theory and Discourse Analysis Theory. The thesis aims to identify the core constituents, form a consensus and enrich the scientific communities’ understanding regarding how these core constituents alongside the discourse contained within the policy documents shape the overall landscape of AI governance in continental Europe. Furthermore, prior to an in depth investigation of the methods and theoretical frameworks mentioned above commences, an introduction is presented to give additional insight to the background of AI & the problem formulation. The results of this study reveal 8 inferred themes. These inferred themes are then thoroughly discussed in alignment with the principles and concepts set forth by the theoretical frameworks. The thesis then provides a conclusive penultimate subchapter that encapsulates the key points and directly addresses the research question before highlighting possible future research opportunities.

This abstract presents a study in which knowledge distillation techniques were applied to a Large Language Model (LLM) to create smaller, more efficient models without sacrificing performance. Three configurations of the RoBERTa model were selected as ”student” models to gain knowledge from a pre-trained ”teacher” model. Multiple steps were used to improve the knowledge distillation process, such as copying some weights from the teacher to the student model and defining a custom loss function. The selected task for the knowledge distillation process was sentiment analysis on Amazon Reviews for Sentiment Analysis dataset. The resulting student models showed promising performance on the sentiment analysis task capturing sentiment-related information from text. The smallest of the student models managed to obtain 98% of the performance of the teacher model while being 45% lighter and taking less than a third of the time to analyze an entire the entire IMDB Dataset of 50K Movie Reviews dataset. However, the student models struggled to produce meaningful results on the topic modeling task. These results were consistent with the topic modeling results from the teacher model. In conclusion, the study showcases the efficacy of knowledge distillation techniques in enhancing the performance of LLMs on specific downstream tasks. While the model excelled in sentiment analysis, further improvements are needed to achieve desirable outcomes in topic modeling. These findings highlight the complexity of language understanding tasks and emphasize the importance of ongoing research and development to further advance the capabilities of NLP models. / Denna sammanfattning presenterar en studie där kunskapsdestilleringstekniker tillämpades på en stor språkmodell (Large Language Model, LLM) för att skapa mindre och mer effektiva modeller utan att kompremissa på prestandan. Tre konfigurationer av RoBERTa-modellen valdes som ”student”-modeller för att inhämta kunskap från en förtränad ”teacher”-modell. Studien mäter även modellernas prestanda på två ”DOWNSTREAM” uppgifter, sentimentanalys och ämnesmodellering. Flera steg användes för att förbättra kunskapsdestilleringsprocessen, såsom att kopiera vissa vikter från lärarmodellen till studentmodellen och definiera en anpassad förlustfunktion. Uppgiften som valdes för kunskapsdestilleringen var sentimentanalys på datamängden Amazon Reviews for Sentiment Analysis. De resulterande studentmodellerna visade lovande prestanda på sentimentanalysuppgiften genom att fånga upp information relaterad till sentiment från texten. Den minsta av studentmodellerna lyckades erhålla 98% av prestandan hos lärarmodellen samtidigt som den var 45% lättare och tog mindre än en tredjedel av tiden att analysera hela IMDB Dataset of 50K Movie Reviews datasettet.Dock hade studentmodellerna svårt att producera meningsfulla resultat på ämnesmodelleringsuppgiften. Dessa resultat överensstämde med ämnesmodelleringsresultaten från lärarmodellen. Dock hade studentmodellerna svårt att producera meningsfulla resultat på ämnesmodelleringsuppgiften. Dessa resultat överensstämde med ämnesmodelleringsresultaten från lärarmodellen.

The key objective of this project was to quantitatively and qualitatively assess the performance of a sentence embedding model, Universal Sentence Encoder (USE), and a word embedding model, word2vec, at the task of topic modelling. The first step in the process was data collection. The data used for the project was podcast descriptions available at Spotify, and the topics associated with them. Following this, the data was used to generate description vectors and topic vectors using the embedding models, which were then used to assign topics to descriptions. The results from this study led to the conclusion that embedding models are well suited to this task, and that overall the USE outperforms the word2vec models. / Det huvudsakliga syftet med det i denna uppsats rapporterade projektet är att kvantitativt och kvalitativt utvärdera och jämföra hur väl Universal Sentence Encoder USE, ett semantiskt vektorrum för meningar, och word2vec, ett semantiskt vektorrum för ord, fungerar för att modellera ämnesinnehåll i text. Projektet har som träningsdata använt skriftliga sammanfattningar och ämnesetiketter för podd-episoder som gjorts tillgängliga av Spotify. De skriftliga sammanfattningarna har använts för att generera både vektorer för de enskilda podd-episoderna och för de ämnen de behandlar. De båda ansatsernas vektorer har sedan utvärderats genom att de använts för att tilldela ämnen till beskrivningar ur en testmängd. Resultaten har sedan jämförts och leder både till den allmänna slutsatsen att semantiska vektorrum är väl lämpade för den här sortens uppgifter, och att USE totalt sett överträffar word2vec-modellerna.

No description available.

No description available.