Abstract In the digital education era, artificial intelligence has developed rapidly in education in China, and hybrid teaching models have become popular. This study aims to explore whether college students’ enjoyment in online classes influences their learning interest and achievement of learning goals, and to examine the moderating role of teacher-student interaction in the realization of learning goals. We conducted an online questionnaire survey of 1736 college students in China to explore the relationships among enjoyment, learning interest, teacher-student interaction, and the achievement of learning goals of online classes. All statistical analyses were performed using SPSS 25.0 and Mplus 8.0. These findings show that college students’ enjoyment influences the achievement of learning goals through the mediating role of learning interest in online classes. The high level of teacher–student interaction is conducive to the transformation of students’ enjoyment into learning interest and the achievement of learning goals. Our study is expected to serve as an important reference for increasing students’ learning interest and achieving their learning goals in online classes.
History of scholarship and learning. The humanities, Social Sciences
Abstract This article investigates the intersection of constructivist principles and statistical methods within Digital Humanities (DH) scholarship. While constructivist principles are foundational, albeit often implicit, in DH scholarship, the increasing use of statistical methods, especially Bayesian approaches, introduces epistemological tensions. Therefore, this article examines the latent constructivist assumptions in DH and explores how these principles resonate with or diverge from the epistemic premises of Bayesian statistics. It argues that there are interferences between Bayesian statistics and constructivism. The main aim is to reveal how epistemological assumptions shape knowledge production, in order to pave the way for a more reflective epistemology of DH scholarship.
Abstract The purpose of this paper is to understand the spatial and temporal evolution of the level of rule of law government construction in China and the mechanism of influence, in an attempt to expand the research direction of legal geography, and to provide empirical cases for how developing countries can promote the rule of law construction under unbalanced geographic, economic and institutional conditions. The study investigates the spatiotemporal variations and influencing factors of law-based government construction levels in each of China’s 31 provincial administrative regions from 2015 to 2022, employing the Moran index and geographic detectors. The results show: (1) The construction level of law-based government in each provincial administrative region has exhibited a clear upward trend, shifting from predominantly “low” and “medium-low” levels to predominantly “medium-high” and “high” levels. (2) The construction level of law-based government and its development type exhibit clear spatial aggregation in each region. The spatial distribution of the four categorized types shows continuity and obvious characteristics of agglomeration. (3) The construction of a law-based government is influenced by economic, social, environmental, and political factors. The basic economic system, level of economic development, and resources per capita available to the administration have the greatest impact. The interaction between these factors significantly enhances their influence on the level of law-based government construction. The improvement of the level of rule of law government construction not only depends on the economic foundation and institutional resources, but is also affected by multiple factors such as urbanization development, demographic structure, public service provision, and institutional innovation path.
History of scholarship and learning. The humanities, Social Sciences
This paper presents the initial findings of a qualitative study investigating the relationship between ethics and education in Italian preschool settings. The research employs the methodology of Constructivist Grounded Theory (CGT) to emphasise the indispensable role of promoting ethical awareness in early childhood education. The preliminary results from the interpretation of intensive interviews with preschool teachers indicate that the structuring of ethically significant educational experiences can be a factor in the growth of ethical awareness. From a theoretical and practical perspective, the study posits that stimulating ethical awareness from early childhood can prevent the rise of an increasingly widespread phenomenon: ethical illiteracy.
Education (General), History of scholarship and learning. The humanities
Abstract This article analyzes the divergence between China’s Personal Information Protection Law (PIPL) and the EU’s General Data Protection Regulation (GDPR), despite their textual similarities. It argues that China’s approach to data protection is shaped by distinct domestic understandings of “risk,” rooted in past legislation, judicial practices, and social concerns. Using focal point theory, the authors identify three key dimensions of risk in China: large-scale participation, economic loss, and threats from third parties. These focal points explain why China’s risk-based approach prioritizes different enforcement goals than the GDPR. The article also shows how these differences manifest in several areas, including the definition of personal information, the regulation of automated decision-making, and the design of enforcement authorities. Ultimately, the article challenges the assumption that legal diffusion through the “Brussels Effect” leads to uniform global standards. Instead, it highlights how domestic cultural and institutional factors reshape transplanted laws, creating seemingly performative enforcement that reflects localized regulatory logics.
History of scholarship and learning. The humanities, Social Sciences
Black-Box Discrete Prompt Learning is a prompt-tuning method that optimizes discrete prompts without accessing model parameters or gradients, making the prompt tuning on a cloud-based Large Language Model (LLM) feasible. Adapting federated learning to BDPL could further enhance prompt tuning performance by leveraging data from diverse sources. However, all previous research on federated black-box prompt tuning had neglected the substantial query cost associated with the cloud-based LLM service. To address this gap, we conducted a theoretical analysis of query efficiency within the context of federated black-box prompt tuning. Our findings revealed that degrading FedAvg to activate only one client per round, a strategy we called \textit{FedOne}, enabled optimal query efficiency in federated black-box prompt learning. Building on this insight, we proposed the FedOne framework, a federated black-box discrete prompt learning method designed to maximize query efficiency when interacting with cloud-based LLMs. We conducted numerical experiments on various aspects of our framework, demonstrating a significant improvement in query efficiency, which aligns with our theoretical results.
Vasileios Sevetlidis, George Pavlidis, Antonios Gasteratos
Positive-Unlabeled (PU) learning addresses classification problems where only a subset of positive examples is labeled and the remaining data is unlabeled, making explicit negative supervision unavailable. Existing PU methods often rely on negative-risk estimation or pseudo-labeling, which either require strong distributional assumptions or can collapse in high-dimensional settings. We propose AngularPU, a novel PU framework that operates on the unit hypersphere using cosine similarity and angular margin. In our formulation, the positive class is represented by a learnable prototype vector, and classification reduces to thresholding the cosine similarity between an embedding and this prototype-eliminating the need for explicit negative modeling. To counteract the tendency of unlabeled embeddings to cluster near the positive prototype, we introduce an angular regularizer that encourages dispersion of the unlabeled set over the hypersphere, improving separation. We provide theoretical guarantees on the Bayes-optimality of the angular decision rule, consistency of the learned prototype, and the effect of the regularizer on the unlabeled distribution. Experiments on benchmark datasets demonstrate that AngularPU achieves competitive or superior performance compared to state-of-the-art PU methods, particularly in settings with scarce positives and high-dimensional embeddings, while offering geometric interpretability and scalability.
Machine learning bias in mental health is becoming an increasingly pertinent challenge. Despite promising efforts indicating that multitask approaches often work better than unitask approaches, there is minimal work investigating the impact of multitask learning on performance and fairness in depression detection nor leveraged it to achieve fairer prediction outcomes. In this work, we undertake a systematic investigation of using a multitask approach to improve performance and fairness for depression detection. We propose a novel gender-based task-reweighting method using uncertainty grounded in how the PHQ-8 questionnaire is structured. Our results indicate that, although a multitask approach improves performance and fairness compared to a unitask approach, the results are not always consistent and we see evidence of negative transfer and a reduction in the Pareto frontier, which is concerning given the high-stake healthcare setting. Our proposed approach of gender-based reweighting with uncertainty improves performance and fairness and alleviates both challenges to a certain extent. Our findings on each PHQ-8 subitem task difficulty are also in agreement with the largest study conducted on the PHQ-8 subitem discrimination capacity, thus providing the very first tangible evidence linking ML findings with large-scale empirical population studies conducted on the PHQ-8.
Radek Trnka, Melisa Schneiderova, Iveta Vojtechova
et al.
SuperAging deserves special attention from researchers in the field of the psychology of aging, because it denotes the preservation of multiple cognitive abilities in very old age. Currently, very little is known about lifestyle factors that could be related to SuperAging. The main goal of the present narrative review was to bring together available evidence involving social factors related to SuperAging and to target avenues for future research. The review summarizes the findings of studies published between 2005 and 2022. Low social participation in midlife age and high social participation in older age were found to be related to SuperAging. In contrast, social network size and diversity did not differ between SuperAgers and cognitively normal older adults. The synthesis of the results indicates that having positive, close, high-quality relationships and a high frequency of social contact may be considered to be hypothetical predictors of superior cognitive performance in later life.
History of scholarship and learning. The humanities, Social Sciences
Abstract This article looks into the implications of urban informality in Katherine Boo’s Behind the Beautiful Forevers: Life, Death and Hope in a Mumbai Slum (2012) as represented in slum life and urban poverty. It aims to investigate the impact of urban poverty on the everyday practices of slum dwellers and their endeavors to escape the trap of poverty in an attempt to highlight the human dimension of the slum. The article seeks to unravel multi-layers of the interaction between people and poverty and the differing models of resistance to poverty and social exclusion depicted in the nonfiction narrative. The article examines slum life from a descriptive sociological perspective with a detailed description of how people survive in poverty. The study of the culture of slums entails an analysis of the survival techniques and everyday practices of slum dwellers, the relations and patterns of behavior among the different categories of people inhabiting a slum, and the outcomes of the interplay between place, culture, and power relations in such communities. This is implemented through an eclectic sociological approach that comprises theories of space, culture, and resistance as proposed by James Scott, Theodore W. Schultz, and Henri Lefebvre.
History of scholarship and learning. The humanities, Social Sciences
This paper aims to examine the literary strategies and political perspectives of Somnuek Chusil, a renowned Nora artist in the Southern Thai region. Contrast to recent international and Thai research on Nora dance that focused on the aspect of ritual, social function, and adaptation under the condition of global modernity, the authors’ analysis centers on the noted Nora artist’s lyrical composition, encompassing 40 of lyrics, to understand his political worldview and literary strategy as part of local experience to the changing of Thai society. The findings reveal four language strategies employed by Somnuek—simile, hyperbole, the use of idioms, and incorporation of the Southern Thai dialect. Of particular note is the dialect’s distinctive role in critiquing political figures and instilling a sense of awareness regarding rights, freedom, and democratic citizenship. Despite the diverse interpretations of political concepts in global academia, Somnuek skillfully harnesses various dialects and writing techniques making him being locally competent interlocutor, and ascending to the status of a famous folk artist in southern Thailand.
We investigate Reinforcement Learning from Human Feedback (RLHF) in the context of a general preference oracle. In particular, we do not assume the existence of a reward function and an oracle preference signal drawn from the Bradley-Terry model as most of the prior works do. We consider a standard mathematical formulation, the reverse-KL regularized minimax game between two LLMs for RLHF under general preference oracle. The learning objective of this formulation is to find a policy so that it is consistently preferred by the KL-regularized preference oracle over any competing LLMs. We show that this framework is strictly more general than the reward-based one, and propose sample-efficient algorithms for both the offline learning from a pre-collected preference dataset and online learning where we can query the preference oracle along the way of training. Empirical studies verify the effectiveness of the proposed framework.
In offline reinforcement learning, in-sample learning methods have been widely used to prevent performance degradation caused by evaluating out-of-distribution actions from the dataset. Extreme Q-learning (XQL) employs a loss function based on the assumption that Bellman error follows a Gumbel distribution, enabling it to model the soft optimal value function in an in-sample manner. It has demonstrated strong performance in both offline and online reinforcement learning settings. However, issues remain, such as the instability caused by the exponential term in the loss function and the risk of the error distribution deviating from the Gumbel distribution. Therefore, we propose Maclaurin Expanded Extreme Q-learning to enhance stability. In this method, applying Maclaurin expansion to the loss function in XQL enhances stability against large errors. This approach involves adjusting the modeled value function between the value function under the behavior policy and the soft optimal value function, thus achieving a trade-off between stability and optimality depending on the order of expansion. It also enables adjustment of the error distribution assumption from a normal distribution to a Gumbel distribution. Our method significantly stabilizes learning in online RL tasks from DM Control, where XQL was previously unstable. Additionally, it improves performance in several offline RL tasks from D4RL.
The current electroencephalogram (EEG) based deep learning models are typically designed for specific datasets and applications in brain-computer interaction (BCI), limiting the scale of the models and thus diminishing their perceptual capabilities and generalizability. Recently, Large Language Models (LLMs) have achieved unprecedented success in text processing, prompting us to explore the capabilities of Large EEG Models (LEMs). We hope that LEMs can break through the limitations of different task types of EEG datasets, and obtain universal perceptual capabilities of EEG signals through unsupervised pre-training. Then the models can be fine-tuned for different downstream tasks. However, compared to text data, the volume of EEG datasets is generally small and the format varies widely. For example, there can be mismatched numbers of electrodes, unequal length data samples, varied task designs, and low signal-to-noise ratio. To overcome these challenges, we propose a unified foundation model for EEG called Large Brain Model (LaBraM). LaBraM enables cross-dataset learning by segmenting the EEG signals into EEG channel patches. Vector-quantized neural spectrum prediction is used to train a semantically rich neural tokenizer that encodes continuous raw EEG channel patches into compact neural codes. We then pre-train neural Transformers by predicting the original neural codes for the masked EEG channel patches. The LaBraMs were pre-trained on about 2,500 hours of various types of EEG signals from around 20 datasets and validated on multiple different types of downstream tasks. Experiments on abnormal detection, event type classification, emotion recognition, and gait prediction show that our LaBraM outperforms all compared SOTA methods in their respective fields. Our code is available at https://github.com/935963004/LaBraM.
Offline Goal-Conditioned RL (GCRL) offers a feasible paradigm for learning general-purpose policies from diverse and multi-task offline datasets. Despite notable recent progress, the predominant offline GCRL methods, mainly model-free, face constraints in handling limited data and generalizing to unseen goals. In this work, we propose Goal-conditioned Offline Planning (GOPlan), a novel model-based framework that contains two key phases: (1) pretraining a prior policy capable of capturing multi-modal action distribution within the multi-goal dataset; (2) employing the reanalysis method with planning to generate imagined trajectories for funetuning policies. Specifically, we base the prior policy on an advantage-weighted conditioned generative adversarial network, which facilitates distinct mode separation, mitigating the pitfalls of out-of-distribution (OOD) actions. For further policy optimization, the reanalysis method generates high-quality imaginary data by planning with learned models for both intra-trajectory and inter-trajectory goals. With thorough experimental evaluations, we demonstrate that GOPlan achieves state-of-the-art performance on various offline multi-goal navigation and manipulation tasks. Moreover, our results highlight the superior ability of GOPlan to handle small data budgets and generalize to OOD goals.
With the increasing demand of individual customption and awareness of cost reduction in express delivery organizations, the Chinese express industry faced with serious challenges especially under the background of government’s strict restrictions on environment and transportation. Therefore, a new service mode called joint distruction (JD) is being tried by the logistics industry, which is expected to address the challenges on online shopping. However, the insufficient understanding of JD adoption factors and their complicated interactions blocks the effectively implementation of the joint distribution. This study aims at identifying potential factors for JD adoption and promoting an effective joint distribution by discovering the interactive relationships among addressed factors. Firstly, potential ingredients for the adoption and implementation of JD are summarized from the literature and industrial interviews. Then, 23 variables are selected and classified into as objectives, drivers, barriers and affected operations. The Interpretive Structural Modeling (ISM) approach is then employed to analyze the crucial factors and the mutual influences amongst 23 variables. Finally, a case study is performed to construct the hierarchical structure of factors toward joint distribution adoption using the proposed ISM-modeling steps. The perplex hierarchical co-relationships are also identified by categorizing the driving variables and dependent variables. Results can assist express enterprises to promote the novel joint distribution mode and acheive higher efficiency of logistics operation by better understanding on crucial factors of JD adoption and implementation.
History of scholarship and learning. The humanities, Social Sciences
تحتاج المکتبات ومراکز المعلومات وکل ماله علاقة بتنظیم وتهیئة المعلومات الى اعادة تقییم بشکل دوری من اجل الوقوف على مستوى الجودة مما یعنی ان النهوض بالواقع العام لهذه المؤسسات بما یضمن رضا کافی من قبل المستفیدین عن الخدمات المقدمة، وهذا ما تم العمل علیه فی هذا البحث اذ تم تطبیق واحد من اهم مقاییس الجودة فی المکتبات ومراکز المعلومات وهو LibQUAL+®) ) فی واحدة من اهم وأقدم المکتبات المرکزیة الجامعیة الا وهی المکتبة المرکزیة لجامعة بغداد بموقعیها الجادریة والوزیریة، وقد بلغت عینة المستفیدین والذین تم توزیع الاستبانة لهم 75 مستفید موزعین على المکتبتین وقد تم استرجاع (68) استمارة منهم، وباستخدام منهج الحالة کوسیلة للوصول لتحقیق اهم الأهداف المرسومة فی هذا البحث ومن ابرزها التوصل الى مستویات جودة الحقیقیة وإیجاد الحلول الملائمة لعلاج جوانب الخلل ان وجدت، إضافة للوصول الى النتائج المرجوة منها والتی کان من أهمها:<br /> - هنالک فجوة حقیقیة فی خدمات المکتبة المقدمة سواء (فجوة اکتفاء) (فجوة تمیز) کون المکتبة تقدم خدمات اقل من المطلوب من قبل المستفید او یتصور انه سیحصل علیها، وهذا اهم هدف حققه المقیاس فی الوصول الى هذه النتیجة.<br /> اما عن اهم التوصیات فکانت:<br /> - الطلب من القیادات الإداریة العلیا للجامعات لإعادة النظر بالدعم الموجه لهذه المؤسسات المهمة من خلال زیادة الدعم سواء المادی ام المعنوی ما یخدم بالنتیجة حرکة البحث العلمی للجامعة التی صارت الیوم احدى رکائز التصنیفات العالمیة لمرتبة الجامعات.
History of scholarship and learning. The humanities