324139 research outputs found
Sort by
Using large language models and speech-to-text models to facilitate the assessment of basic literacy in Ghana
Get PDFThis dissertation examines how recent advances in artificial intelligence, particularly in Natural Language Processing (NLP) and Automatic Speech Recognition (ASR), could enhance the assessment of reading ability in English in West Africa. Although regular diagnostic and formative assessment has proven essential for effective reading instruction, the substantial time and expertise needed to conduct and evaluate reading assessments often restricts their routine implementation, especially in resource-constrained education systems. Recent innovations in Natural Language Processing (NLP) and Automatic Speech Recognition (ASR) offer promising new approaches for making reading assessment more accessible and scalable in low-resource environments.
To investigate this potential, this research develops and analyzes two novel datasets gathered in partnership with Rising Academies, a network of schools in Ghana. The primary dataset encompasses results from various reading assessment tasks administered to 162 students between the ages of 9 and 18, including audio recordings of oral reading, written and spoken responses to comprehension questions, and story retells. The secondary dataset contains high-quality human transcriptions of these same students’ performance on the oral reading fluency tasks, enabling comprehensive analysis of reading errors and fluency patterns. All assessment materials were selected from released items from the 2016 PrePIRLS assessment, which was specifically designed for use in low and middle-income countries. Using these datasets, I then conducted three interconnected empirical studies.
The first study evaluates the viability of using state-of-the-art ASR models to assess oral reading fluency. Using the Ghanaian student dataset, the research determined that the Whisper v2 model could transcribe student reading with high accuracy (Word Error Rate of 10.3%) without any fine-tuning or adaptation. Significantly, the model's performance remained consistent across different student populations, demonstrating similar accuracy when transcribing recordings from both Ghanaian and American students. When these transcriptions were used to calculate Words Correct Per Minute scores, they demonstrated strong agreement with expert human raters (correlation of 0.98).
The second study explores the capacity of Large Language Models to evaluate short answer reading comprehension questions. Drawing from a dataset of over 1,000 student responses, the research established that GPT-4 could achieve accuracy levels matching those of expert human raters. With minimal prompt engineering, the model attained a Quadratic Weighted kappa of 0.91 in the three-class condition and a Linear Weighted kappa of 0.87 in the two-class condition, surpassing existing benchmarks in automated short answer grading.
The third study assesses the effectiveness of LLMs in grading story retell tasks while simultaneously examining the reliability and validity of various retell measures. The research documented strong inter-rater reliability among human raters using both three-class and five-class rubrics (Kendall's W of 0.81 and 0.85 respectively). It also revealed moderate correlations between different measures of retell and other indicators of reading ability, suggesting that retell tasks capture unique aspects of reading comprehension. In evaluating LLM performance, the study found that GPT-4 could effectively replicate human scoring, achieving agreement levels (QWK of 0.82) that nearly matched human-to-human agreement (0.85).
Several themes emerged across these studies. First, the latest generation of AI models exhibited remarkably robust performance across different assessment types without requiring task-specific fine-tuning or extensive prompt engineering. This marks a substantial improvement over previous approaches that demanded considerable technical expertise and task-specific training data. Second, throughout all three studies, the automated approaches reached levels of agreement with human raters that paralleled inter-rater agreement among experts, indicating these technologies could serve as reliable tools for diagnostic and formative assessment purposes. Third, the studies consistently showed minimal evidence of systematic bias in model performance based on student demographics, though this finding warrants careful interpretation given the relatively limited sample sizes.
The research underscores the crucial role of assessment design in successful automation. Specifically, more detailed rubrics that identify meaningful distinctions in student performance appear particularly well-suited to automated scoring. Furthermore, the studies illuminate how different assessment formats may capture distinct aspects of reading ability, reinforcing the importance of employing multiple assessment types to construct a comprehensive picture of student reading proficiency.
These findings contribute to an understanding of how AI can support literacy assessment across diverse educational settings. While previous studies have shown AI's potential for educational assessment in well-resourced environments, this dissertation provides pioneering empirical evidence for the feasibility of applying these technologies to support reading assessment. The consistently strong performance across various assessment types and student populations suggests that recent AI advances could facilitate more regular and comprehensive evaluation of reading ability, particularly in contexts where traditional assessment methods face resource constraints. By demonstrating the viability of automated scoring across multiple assessment types while identifying areas requiring further study, this research enhances our understanding of how artificial intelligence can strengthen literacy assessment in diverse educational contexts
Tucker S. Ferda , Jesus and His Promised Second Coming: Jewish Eschatology and Christian Origins (Grand Rapids, MI: William B. Eerdmans, 2024), pp. xxvi + 538. $69.99
Get PDFInvestigating the mechanisms of membrane protein recognition during protein quality control
No full textMisfolded proteins in the Endoplasmic Reticulum (ER) are degraded by cytosolic proteasomes via a quality control system termed ER-associated protein degradation (ERAD). ERAD consists of multiple branches operating in parallel, defined by a membrane-embedded E3 ubiquitin ligase complex with specificity towards distinct substrates. To date, mechanisms of membrane substrate recognition, as well as the role of additional factors, such as intramembrane proteases, in this process, remain elusive.
Recent genome-wide CRISPR/Cas9 genetic screen for ERAD components implicated in the clearance of a model misfolded membrane protein, CYP51A1TM, identified a novel ERAD branch comprised of the RNF185/Membralin complex. Building on these initial observations, this thesis presents a mechanistic insight into quality control factors involved in membrane protein recognition by ERAD, using a range of interdisciplinary approaches. First, I have conducted a comparative analysis of simplified model substrates, derived from twenty human ER-membrane resident cytochrome P450 enzymes and investigated features that drive substrate recognition by the E3 ubiquitin ligase complexes. In addition, I have identified an aspartyl intramembrane protease, Signal Peptide Peptidase (SPP), as a non-canonical protein quality control factor involved in recognition and clearance of the human lanosterol demethylase CYP51A1, upstream of the RNF185/MBRL complex. Finally, we have discovered a novel substrate of the RNF185/MBRL complex and characterised the key determinants of its recognition by ERAD. This thesis further advances the knowledge on remarkable specificity of distinct ERAD branches, pinpointing determinants that drive membrane substrate recognition and selection for quality control
Color symmetry and altermagneticlike spin textures in noncollinear antiferromagnets
Get PDFWe present a formalism based on colour symmetry to analyse the momentum-space spin textures of non-collinear antiferromagnets. We show that, out of the spin textures allowed by the magnetic point group, . We demonstrate this approach in the case of three complex, non-collinear magnets, Mn3Ir(Ge,Si), Pb2MnO4 and Mn3GaN. For Mn3GaN, we also show that the predictions of colour-symmetry analysis are consistent with density functional theory calculations performed on the same system both with and without spin-orbit coupling
The welfare and market effects of delays in humanitarian assistance
Get PDFDelays in aid delivery are common, yet their impacts on households and markets are theoretically ambiguous and empirically understudied. Models with financial constraints or present bias predict sharp consumption declines, while the Permanent Income Hypothesis predicts consumption smoothing. We test these predictions using high-frequency data and random interview timing in a large refugee camp in Kenya. Households smooth consumption under regular aid cycles, but delays reduce food consumption and food security. Informal credit through shops mitigates short-term impacts, but entails higher prices (+17%). Prices also respond to the timing of aid. Results support credit constraint models
WERF Endometriosis Phenome and Biobanking Harmonisation Project for Experimental Models in Endometriosis Research (EPHect-EM-Pain): methods to assess pain behaviour in rodent models of endometriosis
Get PDFPain is a debilitating symptom of endometriosis, and its mechanisms are often explored using rodent models. However, a lack of harmonization amongst models and behavioural measures, in addition to inconsistent reporting, might limit the overall clinical relevance and hinder translation of findings. An additional challenge is accurately linking rodent behaviour to human experiences of endometriosis. This study aimed to: (i) review current measures of pain-associated behaviours used in endometriosis studies; (ii) recommend best practices for each method and their suitability to study endometriosis-associated pain; and (iii) develop internationally agreed-upon standard operating procedures (‘EPHect-EM-Pain SOPs’). The World Endometriosis Research Foundation (WERF) assembled an international working group, from which a ‘pain behaviour working group’ consisting of experts in the field was established. The group used additional consultation from experimental pain model scientists in the broader field. Stimulus-evoked (reflexive) and stimulus-independent (spontaneous) measures are currently used to assess pain-associated behaviours in rodents with experimental endometriosis. All existing methods offer advantages and limitations regarding ethological relevance, output quality, and equipment/training requisites. Internationally standardized pain SOPs as well as summary documentation outlining the minimum and standard requirements for several behavioural measures were developed, as well as consensus recommendations on experimental designs and documentation. To more closely reflect the lived experiences of those with endometriosis, the consortium recommends that, following validation, multiple types of pain-related and/or parallel rodent behaviours (e.g. anxiety) should be quantified as surrogate outcome measures for endometriosis-associated pain. These harmonized methods and documentation for endometriosis research will facilitate essential comparisons among studies, improve translational applicability, and provide a superior holistic view of animal (and thus human) wellbeing
What motivate consumers’ purchase intention and the intention to continue watching in livestream shopping
Get PDFA growing number of companies are adopting livestream shopping as a means of recommending their products to consumers. However, the question of whether the purchase intention for sustainable clothing amongst consumers can be increased by livestream shopping remains unclear. Therefore, the present study was designed to clarify the relationship between cognitive reactions (vividness, attractiveness, flow, and multisensory cues), emotion (arousal, pleasure), and behaviour (intention to continue watching, purchase intention) during livestream shopping for sustainable clothing. Factor analysis and structural equation model are adopted to analyse the data. The effect of social sharing, ‘stickiness’, and social presence are also explored in a research model. In this study, people’s cognitive reactions were found to have positive impacts on the emotion that consumers associate with sustainable clothing in the context of livestreaming shopping. Moreover, both arousal and pleasure mediated the relationship between cognitive reactions and behavioural intention, while stickiness mediated the relationship between cognitive reactions and purchase intention, and social presence positively impacts people’s intention to continue watching. Social sharing significantly affects purchase intention. These results therefore provide brands and companies with a number of actionable insights to adopt appropriate marketing strategies in the context of livestream shopping
Signs of authority and sites of spirits: strange trees in Early and Medieval China
Get PDFThis thesis looks at representations of unusual trees in collections of accounts of anomalies in pre-Tang China. These texts were collected in various sources, including collections dedicated to anomalies, sections of official histories, incidentally mentioned in biographies of notable figures, and encyclopaedic compilations. By examining representations of anomalous natural phenomena in Classical Chinese texts, this study will attempt to reconstruct pre-modern Chinese knowledge of the natural world. In these texts, plant behaviour that would be regarded as impossible (or at least highly unlikely) in science signified something about the human world, such as proper or improper governance. They also were collected and recorded as part of larger regimes of knowledge, such as the political and spiritual power represented by the figure of the emperor and by religious traditions like Buddhism and Daoism. This connection between unusual plant behaviour and good human behaviour can also be seen in accounts of trees planted on tombs, which responded to the idealised qualities of their occupants. However, despite this connection to the human world, strange plants most commonly behaved in plant-like ways, and when depicted with human-like interiority, usually were described as being occupied by human or animal-like entities. This thesis shows that even when described as unusual, plants were still understood in relation to humans, but rather than therefore being regarded as essentially similar to humans, they nevertheless retained their distinctive plant-like qualities
The In-Plane Compression Response of Thermoplastic Composites: Effects of High Strain Rate and Type of Thermoplastic Matrix
Get PDFDesigning thermoplastic composites for particular uses requires understanding their dynamic mechanical behaviour, which affects how well they operate in practical settings. The Split Hopkinson pressure bar (SHPB) test allows for evaluating these materials’ responses to high strain rates. In this study, an in-situ laser-assisted fibre placement (LAFP) machine has been utilised to produce laminate composites with varied designs, i.e., different angles of layers [0/45/–45/90]4s, using three types of thermoplastic tapes (UD-CF/PPS, UD-CF/PEEK, and UD-CF/PEKK). Using a servo-hydraulic testing machine and SHPB apparatus, we have examined the dynamic compressive behaviour of thermoplastic laminate composites with various matrices (PPS, PEEK, and PEKK) in in-plane directions and at strain rates of approx. 0.001, 0.1, 10, 800, 1800/s. Experimental results indicate that the type of thermoplastic matrix and strain rate significantly affect how the laminate composites behave. The in-plane compressive strength and modulus increase approximately linearly with the strain rate. According to the fracture of morphological pictures, the main failure mechanism of all three types of specimens is shear failure under in-plane compression loads, which is followed by delamination and burst
Extracellular Vesicles as Targeted Communicators in Complementary Medical Treatments
Get PDFThe supposed meridians of traditional oriental medicine have been a cause of conflict between traditional and modern medical science. A possible resolution has been proposed: That extracellular vesicles, including exosomes, may be the transmitters of traditional therapies such as massage and acupuncture. This article develops that idea by proposing that the pathways between surface and deep structures may be laid down during the embryonic migration of cells from one region of the developing body to distant regions. This hypothesis depends on the proven targeting of vesicular communication via cell surface binding molecules and their complementary binding sites on target cells. The hypothesis is therefore experimentally testable. The article also draws attention to a strong analogy with Charles Darwin’s theory of pangenesis for particulate communication between the soma and germline