Maximum deviation global discrimination index (MDGDI) is a new item selection method for cognitive diagnostic computerized adaptive testing that allows for attribute coverage balance. We developed the maximum limitation global discrimination index (MLGDI) from MDGDI, which allows for both attribute coverage balance and item exposure control. MLGDI can realize the attribute coverage balance and exposure control of the item. Our simulation study aimed to evaluate the performance of our new method against maximum global discrimination index (GDI), modified maximum GDI (MMGDI), standardized weighted deviation GDI (SWDGDI), and constraint progressive with SWDGDI (CP_SWDGDI). The results indicated that (1a) under the condition of realizing the attribute coverage balance, MDGDI had the highest attribute classification accuracy; (1b) when the selection strategy accommodated the practical constraints of the attribute coverage balance and item exposure control, MLGDI had the highest attribute classification accuracy; (2) adding the item exposure control mechanism to the item selection method reduces the classification accuracy of the attributes of the item selection method; and (3) compared with GDI, MMGDI, SWDGDI, CP_SWDGDI, and MDGDI, MLGDI can better achieve the attribute-coverage requirement, control item exposure rate, and attribute correct classification rate.

This paper explores Knowledge Tracing (KT) and its role in modeling students’ knowledge progression over time. While traditional approaches such as Bayesian Knowledge Tracing and Deep Knowledge Tracing have established a solid foundation in this field, previous research often overlooked the complex interactions between problem features and learner behaviors. To address this gap, this paper introduces the Knowledge Tracking Model for Integrating Question Information and Learning Behaviour (IQILB-KT), a novel framework that enhances the precision of knowledge tracing in online educational systems. By integrating the effects of question characteristics and student behaviour via a channel attention mechanism, the model captures deeper insights into both question information and the learning process. Furthermore, the IQILB-KT model employs a Bottleneck Attention Module (BAM) to calculate learning behaviour synergies, replacing the previous reliance on covariance statistics. This enhances the stability and accuracy of the model when dealing with learning behaviour synergy data. Finally, the efficacy of the IQILB-KT model was validated through rigorous testing on two real-world datasets from the ASSISTments educational platform. The results demonstrate that the model outperforms all existing methods in predicting students’ future performance and tracking their knowledge states in real time.

Given the continued relevance of perceived scarcity, understanding how a scarcity mindset influences human cooperation remains critical. However, previous research has yielded mixed results regarding this relationship. To clarify these inconsistencies, this study examined the impact of a scarcity mindset on cooperation within various social normative contexts. Participants were randomly assigned to either a scarcity or abundance mindset and engaged in a Public Goods Game under descriptive or injunctive normative conditions, each with high or low prosocial expectations. The results revealed that in both normative contexts, individuals with a scarcity mindset exhibited greater cooperation than those with an abundance mindset. Individuals also cooperated more under high prosocial norms compared to low ones. Importantly, the interaction effect revealed that while cooperation did not differ between the two mindsets under low prosocial norms, the scarcity mindset led to significantly greater cooperation under high prosocial norms. These findings provide new insights into the scarcity–cooperation dynamic and suggest that robust prosocial norms can amplify the cooperative tendencies associated with a scarcity mindset, highlighting the importance of leveraging social norms to enhance cooperation in resource-scarce situations.

Small-scale (e.g., classroom) assessment represents the most common and needed scenario for cognitive diagnostic testing. In such settings, polytomously scored items (e.g., constructed-response tasks) are widely used, as they provide more fine-grained measurement of students’ skills and cognitive processes. However, a significant gap remains between the current methods and pressing practical needs. On one hand, parametric cognitive diagnosis models capable of handling polytomous response data require large samples for stable estimation, making them unsuitable for small-scale classroom use. On the other hand, existing nonparametric classification methods, while robust in small samples, are largely confined to dichotomous (0/1) response data. There is a lack of dedicated nonparametric methods for polytomous responses, creating a disconnect between practical testing and diagnostic tools. To address this real-world necessity, this study proposes the seq-GNPED method. It extends the generalized nonparametric classification framework to polytomous data by introducing weighted ideal category response and a collapsed class iterative algorithm. Simulations and empirical applications confirm that seq-GNPED achieves robust and accurate diagnosis under small sample conditions where parametric models falter, effectively leveraging the informational richness of polytomous items. This work bridges a critical gap by providing a practical, nonparametric tool tailored for fine-grained, classroom-ready cognitive diagnosis.

In educational measurement, exploring the method of generating multiple high-quality parallel tests has become a research hotspot. One purpose of this research is to construct parallel forms item by item according to a seed test, using two proposed item selection heuristic methods [minimum parameters–information–distance method (MPID) and minimum information–parameters–distance method (MIPD)]. Moreover, previous research addressing test assembly issues has been limited mainly to situations in which the information curve of the item pool or seed test has a normal or skewed distribution. However, in practice, the distributions of information curves for tests are diverse. These include multimodal distributions, the most common type of which is the bimodal distribution. Therefore, another main aim of this article is to extend the information curves of unimodal distributions to bimodal distributions. Thus, this study adopts simulation research to compare the results of two item, response, theory (IRT)-based item matching methods (MPID and MIPD) using different information curve distributions for item pools or seed tests. The results show that the MPID and MIPD methods yield rather good performance in terms of both two statistical targets when the information curve has a unimodal distribution, and two new methods yield better performance than two existing methods in terms of test information functions target when the information curve has a bimodal distribution.

This paper proposes a novel framework, HyperGAT-BERT-RAS, that integrates: (1) a Hy perGraph Attention Network (HyperGAT) with BERT for enhanced semantic representa-tion; (2) a Reference Answer Set (RAS) constructed via clustering of full-score answers; (3) Siamese Neural Networks (SNNs) for similarity-based scoring; and (4) GPT-4-based data augmentation to address class imbalance. Experiments on the Ohsumed and ASAP-5 da-tasets demonstrate that: (i) HyperGAT-BERT achieves 0.7317 accuracy on Ohsumed text classification, outperforming baseline HyperGAT by 2.69%; (ii) the full Hyper-GAT-BERT-RAS achieves 0.7991 accuracy and 0.7956 F1-score, with RAS contributing the most to performance gains (4.34% accuracy drop when removed); (iii) GPT-4 augmentation improves Quadratic Weighted Kappa from 0.584 to 0.880 and minority-class (scores 2–3) F1 by 15.3%. These improvements translate into more reliable scoring of diverse student answers, reduced teacher grading burden, and enhanced feasibility of AI-assisted forma-tive assessment in real classrooms. Ablation and error analyses confirm the contribution of each component. The framework advances ASAG by synergizing graph-based relational modeling, pretrained language understanding, and knowledge-guided scoring.