Educational games have been increasingly used to improve students’ computational thinking. However, most existing games have focused on the theoretical knowledge of computational thinking, ignoring the development of computational thinking skills. Moreover, there is a lack of integration of adaptivity into educational computer games for computational thinking, which is crucial to addressing individual needs in developing computational thinking skills. In this study, we present an adaptive educational computer game, called AutoThinking, for developing students’ computational thinking skills in addition to their conceptual knowledge. To evaluate the effects of the game, we conducted an experimental study with 79 elementary school students in Estonia, where the experimental group learned with AutoThinking, while the control group used a traditional technology-enhanced learning approach. Our findings show that learning with the adaptive educational computer game significantly improved students’ computational thinking related to both conceptual knowledge and skills. Moreover, students using the adaptive educational computer game showed a significantly higher level of interest, satisfaction, flow state, and technology acceptance in learning computational thinking. Implications of the findings are also discussed.

A significant amount of research has indicated that students’ procrastination tendencies are an important factor influencing the performance of students in online learning. It is, therefore, vital for educators to be aware of the presence of such behavior trends as students with lower procrastination tendencies usually achieve better than those with higher procrastination. In the present study, we propose a novel algorithm—using student’s assignment submission behavior—to predict the performance of students with learning difficulties through procrastination behavior (called PPP). Unlike many existing works, PPP not only considers late or non-submissions, but also investigates students’ behavioral patterns before the due date of assignments. PPP firstly builds feature vectors representing the submission behavior of students for each assignment, then applies a clustering method to the feature vectors for labelling students as a procrastinator, procrastination candidate, or non-procrastinator, and finally employs and compares several classification methods to best classify students. To evaluate the effectiveness of PPP, we use a course including 242 students from the University of Tartu in Estonia. The results reveal that PPP could successfully predict students’ performance through their procrastination behaviors with an accuracy of 96%. Linear support vector machine appears to be the best classifier among others in terms of continuous features, and neural network in categorical features, where categorical features tend to perform slightly better than continuous. Finally, we found that the predictive power of all classification methods is lowered by an increment in class numbers formed by clustering.

As education is classified as a ’high-risk’ domain under regulatory frameworks like the EU AI Act, ensuring accurate interpretability in AI applications is crucial due to their significant impact on students’ outcomes and future prospects. This study compares a knowledge-based artificial neural network (KBANN) with a conventional artificial neural network (ANN) in educational predictive modelling, focusing on generalizability, interpretability, and the fidelity of post-hoc explanations. Our findings reveal that while both models exhibit similar performance, the KBANN distinctly incorporates structured educational knowledge, aligning more closely with key educational principles and causal relationships. Post-hoc explanation methods, including Kernel SHAP, Permutation SHAP, and LIME, are used to elucidate the ANN’s decision-making processes, revealing considerable variability in interpreting feature importance. The fidelity and reliability of these explanation methods are further scrutinized through experiments simulating predictions based on the KBANN’s rules. The results indicate that while some methods, like LIME, show slightly better alignment with KBANN in specific performance metrics, overall, they fall short in replicating the underlying decision-making process of KBANN, as evidenced by low correlation coefficients. Furthermore, global feature importance analysis shows that post-hoc methods often fail to accurately reflect the structured knowledge integrated by KBANN, misattributing importance to less relevant features. This misalignment and the resultant discrepancies in feature interpretation challenge the reliability of these explanations, suggesting they may not provide a trustworthy or accurate basis for understanding the predictive models. Lastly, KBANN achieved significantly lower computation times, highlighting the practical efficiency of this method. These findings collectively underscore the limitations of post-hoc explanation methods in conveying the true reasons behind model predictions, urging caution among stakeholders and researchers when relying on these methods for interpretative insights in educational contexts.

Artificial neural networks (ANNs) have proven to be among the most important artificial intelligence (AI) techniques in educational applications, providing adaptive educational services. However, their educational potential is limited in practice due to challenges such as the following: (i) the difficulties in incorporating symbolic educational knowledge (e.g., causal relationships and practitioners’ knowledge) in their development, (ii) a propensity to learn and reflect biases, and (iii) a lack of interpretability. As education is classified as a ‘high-risk’ domain under recent regulatory frameworks like the EU AI Act—highlighting its influence on individual futures and discrimination risks—integrating educational insights into ANNs is essential. This ensures that AI applications adhere to essential educational restrictions and provide interpretable predictions. This research introduces NSAI, a neural-symbolic AI approach that integrates neural networks with knowledge representation and symbolic reasoning. It injects and extracts educational knowledge into and from deep neural networks to model learners’ computational thinking, aiming to enhance personalized learning and develop computational thinking skills. Our findings revealed that the NSAI approach demonstrates better generalizability compared to deep neural networks trained on both original training data and data enriched by SMOTE and autoencoder methods. More importantly, we found that, unlike traditional deep neural networks, which mainly relied on spurious correlations in their predictions, the NSAI approach prioritizes the development of robust representations that accurately capture causal relationships between inputs and outputs. This focus significantly reduces the reinforcement of biases and prevents misleading correlations in the models. Furthermore, our research showed that the NSAI approach enables the extraction of rules from the trained network, facilitating interpretation and reasoning during the path to predictions, as well as refining the initial educational knowledge. These findings imply that neural-symbolic AI not only overcomes the limitations of ANNs in education but also holds broader potential for transforming educational practices and outcomes through trustworthy and interpretable applications.

Despite an increasing consensus regarding the significance of properly identifying the most suitable clustering method for a given problem, a surprising amount of educational research, including both educational data mining (EDM) and learning analytics (LA), neglects this critical task. This shortcoming could in many cases have a negative impact on the prediction power of both the EDM and LA based approaches. To address such issues, this work proposes an evaluation approach that automatically compares several clustering methods using multiple internal and external performance measures on 9 real-world educational datasets of different sizes, created from the University of Tartu’s Moodle system, to produce two-way clustering. Moreover, to investigate the possible effect of normalization on the performance of the clustering algorithms, this work performs the same experiment on a normalized version of the datasets. Since such an exhaustive evaluation includes multiple criteria, the proposed approach employs a multiple criteria decision-making method (i.e., TOPSIS) to rank the most suitable methods for each dataset. Our results reveal that the proposed approach can automatically compare the performance of the clustering methods and accordingly recommend the most suitable method for each dataset. Furthermore, our results show that in both normalized and nonnormalized datasets of different sizes with 10 features, DBSCAN and k-medoids are the best clustering methods, whereas agglomerative and spectral methods appear to be among the most stable and highly performing clustering methods for such datasets with 15 features. Regarding datasets with more than 15 features, OPTICS is among the top-ranked algorithms among the nonnormalized datasets, and k-medoids is the best among the normalized datasets. Interestingly, our findings reveal that normalization may have a negative effect on the performance of certain methods, e.g., spectral clustering and OPTICS; however, it appears to mostly have a positive impact on all of the other clustering methods.

Pre-processing and post-processing are significant aspects of natural language processing (NLP) application software. Pre-processing in neural machine translation (NMT) includes subword tokenization to alleviate the problem of unknown words, parallel corpus filtering that only filters data suitable for training, and data augmentation to ensure that the corpus contains sufficient content. Post-processing includes automatic post editing and the application of various strategies during decoding in the translation process. Most recent NLP researches are based on the Pretrain-Finetuning Approach (PFA). However, when small and medium-sized organizations with insufficient hardware attempt to provide NLP services, throughput and memory problems often occur. These difficulties increase when utilizing PFA to process low-resource languages, as PFA requires large amounts of data, and the data for low-resource languages are often insufficient. Utilizing the current research premise that NMT model performance can be enhanced through various pre-processing and post-processing strategies without changing the model, we applied various decoding strategies to Korean–English NMT, which relies on a low-resource language pair. Through comparative experiments, we proved that translation performance could be enhanced without changes to the model. We experimentally examined how performance changed in response to beam size changes and n-gram blocking, and whether performance was enhanced when a length penalty was applied. The results showed that various decoding strategies enhance the performance and compare well with previous Korean–English NMT approaches. Therefore, the proposed methodology can improve the performance of NMT models, without the use of PFA; this presents a new perspective for improving machine translation performance.