Intelligent Technology for Educational Applications

Table of Contents

1. Frontmatter

2. AI-Driven Personalized Learning and Adaptive Systems

   1. Frontmatter

   2. Markov Chain Monte Carlo Methods for Dynamic Student Assessment

      Aled Williams

      Abstract

      In this paper we explore an application of Markov Chain Monte Carlo (MCMC) methods to the field of educational assessment. The aim of our approach is to dynamically evaluate student understanding and adaptively tailor question difficulty and topic coverage in real-time. In particular, our model employs a variant of a classical algorithm to decide on question transitions, optimising the assessment process to balance exploration of the student’s knowledge base and exploitation of known learning gaps. The model not only adjusts the sequence of questions based on previous responses, but also integrates student-reported confidence levels to refine the estimation of their understanding.

   3. An Embedded Generative AI Agent for Personalised Documentation in Early Childhood Education

      Wenjun Liu, Chenwei Liang, Shiyu Sun, Shengjun Ma

      Abstract

      Generative AI has been increasingly explored across various domains, including education. However, its application in early childhood education (ECE) for assisting teachers’ administrative tasks remains limited. In this study, an embedded generative AI agent is developed and deployed on the Jetson Orin Nano platform, integrated with a touchscreen interface and an onboard camera for data input. Log generation is performed entirely on-device through simple user inputs and real-time image analysis, enabling automated creation of personalised child activity logs. The system's effectiveness is evaluated in terms of accuracy and efficiency. In addition, data privacy concerns associated with collecting children’s behavioural data are addressed, offering insights into the practical deployment of embedded AI agents in ECE environments.

   4. Effective Educational Decisions Through Neural Prediction

      Emily Opoku Aboagye-Dapaah, Leo Paapa Tattrah, Sylvester Agyen, Joshua Caleb Dagadu

      Abstract

      This project utilizes a Multi-Layer Perceptron (MLP) Neural Network to predict the academic performance of students based on demographic information, study habits absenteeism and parental education levels for their text scores. The MLP model was trained using students’ records, which underwent pre-processing steps including normalization and categorical encoding to prepare the data for analysis. The model was evaluated based on its ability to predict students’ Grade Point Average (GPA), using performance metrics such as Mean Squared Error (MSE) and R-squared (R²). The model achieved an R² score of 0.71, indicating that it successfully captured a significant portion of the variability in academic performance. Key predictors of academic performance included study hours, test scores, and absenteeism, while the model encountered challenges in predicting outcomes for students with irregular academic behaviours. This research demonstrates the potential of Multi-Layer Perceptron neural networks in predicting student academic outcomes and suggests future improvements by incorporating larger datasets and more diverse features to enhance prediction accuracy and applicability.

   5. Research on Dynamic Evaluation and Feedback Algorithm of University Teachers' Digital Literacy Based on Multimodal Data Fusion

      Yu Guo

      Abstract

      As a key indicator of teachers’ teaching ability in digital environment, digital literacy covers many aspects such as information acquisition, processing, analysis and innovative application. The traditional evaluation system often ignores this comprehensive index and adopts a single quantitative or qualitative model, which is difficult to fully reflect the true performance of teachers. In this study, multimodal data fusion technology is introduced to integrate multimodal data such as teaching videos, teaching logs, student feedback, etc., features are extracted and fused by deep learning algorithm, and weights are dynamically adjusted by reinforcement learning to realize real-time and comprehensive evaluation of teachers’ digital literacy. The experimental results show that the accuracy of the fused features is significantly improved, and the scores of teachers’ digital literacy are highly correlated with the actual performance, and the scores show an upward trend with time, reflecting the dynamic improvement of teachers’ digital literacy. In addition, the personalized feedback report has been highly recognized by teachers, effectively helping teachers improve their digital literacy. Compared with the traditional evaluation methods, this algorithm is excellent in accuracy, comprehensiveness and dynamics, which provides a new scientific basis and method for the evaluation of university teachers’ digital literacy, and also provides a reference for multimodal data fusion and evaluation in other fields.

   6. Variations in University Educators’ Teaching Experiences with Generative AI

      Feifei Han, Robert Ellis, Henry Cook

      Abstract

      Generative artificial intelligence (GenAI) is becoming more embedded across various sectors, including tertiary education. Consequently, universities are under increasing pressure to equip students with the skills necessary to engage with GenAI in their future professions. While GenAI technologies, teaching methods, and ethical debates continue to evolve, there is still limited understanding of how university educators practically use GenAI in their teaching. This study uses a phenomenographic method to investigate how faculty members perceive and integrate GenAI into their instructional approaches. Based on semi-structured interviews with 30 educators, the research uncovers a range of perspectives and applications of GenAI in the classroom. These differing views reveal significant qualitative differences that deepen our understanding of how educators are responding to GenAI in academic settings. The findings indicate that certain perceptions and pedagogical practices are more effective in helping students develop the competencies needed in today’s workforce. Moreover, the study provides insights that can guide academic leaders in creating meaningful support systems for faculty employing GenAI as an educational tool. By focusing on educators’ experiences, the research adds valuable knowledge to the broader discussion on GenAI’s impact on teaching and learning in tertiary education.

   7. Exploring the Potential of GPT-4o’s Vision in Vector Visualisation for Pre-tertiary Mathematics

      Nguyen Thanh Minh Le, Chanoudam Sopheap, Kenneth Y. T. Lim

      Abstract

      This research project delves into the potential of machine learning and computer vision to assist students in visualising complex mathematical concepts, with a specific focus on vectors. Recognising the challenges students face in grasping abstract mathematical notions, the study investigates the development of a system using GPT-4o that automates the visualisation of handwritten vector equations. The system employs computer vision, a rapidly evolving field in artificial intelligence, to analyse and interpret handwritten mathematical expressions. By extracting key components such as variables, coefficients, operators, and vector notation, the system generates precise graphical representations of vector equations. This approach aims to bridge the gap between theoretical concepts and visual understanding, enabling students to develop a more intuitive grasp of vectors. The research evaluates the system’s accuracy in recognising and interpreting diverse handwriting styles, assessing its potential to improve students’ understanding of vector concepts. While acknowledging challenges such as the need for extensive training datasets and refined algorithms to handle the nuances of handwritten mathematical notation, the project outlines future work including expanding the system’s scope to other mathematical topics, incorporating interactive features, and integrating with augmented reality for a more immersive experience. The ultimate goal is to seamlessly integrate this tool into students’ learning devices, providing accessible and convenient support for their mathematical development. This research contributes to the advancement of AI-driven educational tools, showcasing the potential of machine learning to enhance students’ learning experiences and bridge the gap between abstract concepts and visual understanding.

   8. Designing an AI-Supported Pedagogical Evaluation Model for Visual and Illustration Design in China

      Lin Yu, Gening Liu, Ziwei Yu

      Abstract

      PURPOSE: This study examines the impact of AI-supported feedback on learning effectiveness and design quality in visual and illustration design education in China, and explores the moderating role of perceived usefulness of AI tools. METHOD: A quantitative, survey-based approach was used with data from 430 university students. Stratified random sampling and power analysis determined the sample size. Data were analysed using descriptive statistics and SEM via Smart PLS. FINDINGS: AI-supported feedback significantly enhances learning effectiveness and design quality. Perceived usefulness moderates these effects, with students’ recognition of AI’s value influencing outcomes. ORIGINALITY: The study provides empirical support for integrating AI feedback in design education, offering insights to improve pedagogy, engagement, and creative outcomes.

3. Intelligent Tools for Language Learning and Translation

   1. Frontmatter

   2. The Dual Impact of Speech Recognition Technology on Simultaneous Translation

      Maria Borodina

      Abstract

      Currently, technologies such as artificial intelligence, big data, speech recognition technology, and machine translation are having an increasing impact on the field of human translation. On the one hand, they challenge manual translation, but on the other, they help and improve it. This article analyzes the positive and negative effects of speech recognition technology in simultaneous translation on the interpretation process, examines and analyzes the results of previous studies, and provides some recommendations to other translators on the use of speech recognition technology in simultaneous translation.

      The author emphasizes that speech recognition technology has a significant impact on the simultaneous translation process, improving its quality and efficiency. The dual impact of this technology on translation synchronization is, on the one hand, in automating routine operations, reducing the burden on the translator, and on the other, in making work more difficult, increasing the requirements for reaction speed and accuracy. Automation of text input processes using voice technology makes it possible to transfer source speech into the target language faster, minimizing delays between speakers and listeners. However, in order to work successfully, translators have to adapt to new conditions, promptly correcting machine recognition results in order to avoid mistakes.

      The study examines the possibilities and limitations of modern speech recognition systems in relation to simultaneous translation, identifying key success factors for integrating technology into this process. Special attention is paid to the interaction of the human factor and algorithms, which determines the future development of synchronous interpretation. The work highlights the need to train specialists who are able to effectively use modern tools while maintaining high-quality communication.

   3. Special Features of English-Russian Translation of Scientific Texts Based on Artificial Intelligence

      Maria Borodina, Elena Monakhova, Irina Gorofonova

      Abstract

      The translation of scientific texts from English into Russian presents unique challenges due to linguistic, cultural, and terminological differences between the two languages. This study explores the special features of English-Russian translation in the context of artificial intelligence applications. The research focuses on identifying key factors that influence the accuracy and quality of machine-generated translations for scientific content. These include issues such as domain-specific vocabulary, syntax complexity, and contextual understanding. By analyzing both traditional human-driven approaches and AI-based methods, this work highlights areas where machine learning algorithms can enhance or struggle with precision when translating highly technical material. The findings aim to provide insights for improving automated translation systems while addressing the nuances required for accurate rendering of scientific discourse across languages.

   4. The Use of Multimedia in Self-directed Pronunciation Learning: Vietnamese EFL Learners’ Perspectives

      Thi Duyen Phuong, Thi Thanh Huyen Phuong

      Abstract

      Despite the growing potential of multimedia tools in assisting different aspects of English learning, limited research explores students’ self-directed pronunciation learning using multimedia, especially in under-resourced contexts such as Vietnam. Addressing this gap, his study investigates university students’ use of multimedia tools for English pronunciation learning within a self-directed framework. The study finds that although students frequently engaged with multimedia for general English language acquisition, pronunciation was not the primary focus. However, over half preferred multimedia tools over traditional resources, highlighting convenience, multi-sensory engagement, and accessibility—especially via smartphones. Platforms like YouTube, Facebook, and TikTok were most popular, while dedicated apps like ELSA Speak remained underutilized, primarily due to cost. While most participants found multimedia engaging and motivational, many reported limited awareness of how to effectively leverage these tools for systematic pronunciation improvement. Challenges included difficulty identifying appropriate resources, limited teacher guidance, and financial constraints. Despite moderate confidence levels, students expressed a strong preference for autonomous learning supplemented by clearer instructional support. Findings underscore the need for pedagogical scaffolding to bridge informal multimedia usage with targeted pronunciation learning outcomes in digital contexts.

   5. Hybrid Learning in English Education: Perceptions of Vietnamese Junior High School Teachers

      Thi Duyen Phuong, Thi Thanh Huyen Phuong

      Abstract

      As hybrid learning continues to gain momentum globally, understanding how it is perceived by educators is essential for its effective integration into general education. This study investigates the perceptions of junior high school English teachers in northern Vietnam regarding the benefits, challenges, and institutional support associated with hybrid learning. A total of 384 teachers from diverse school types (public and private), geographic locations (urban, suburban, and rural), and age groups participated in a structured survey. Data were analyzed using SPSS, employing reliability checks and inferential statistics, including ANOVA and t-tests. Findings indicate that teachers generally perceive hybrid learning as beneficial for promoting personalized instruction, student autonomy, and enhanced access to digital resources. Significant differences were observed across age groups. Also, urban teachers and those aged 55 and above identified greater implementation challenges, often related to complexity and technological adaptation and students’ readiness. Additionally, public school teachers reported significantly higher levels of institutional support compared to their private school counterparts.

      The study highlights the need for differentiated professional development, targeted policy support, and equitable infrastructure to ensure the effective implementation of hybrid learning. It calls for the formal incorporation of hybrid learning models into Vietnam’s national curriculum to align with global pedagogical shifts in the post-pandemic context.

   6. The Application of Deep Learning-Based Intelligent Translation Technique Training and Evaluation System in French Language Teaching

      Wen Pu

      Abstract

      Recently, the rapid development of artificial intelligence (AI) and computer technology has promoted the widespread application of intelligent translation systems in various industries, demonstrating enormous potential and important value. In the study and practice of French, the role of translation is irreplaceable. However, traditional French translation teaching is limited by resources, methods, and evaluation tools, making it difficult to meet the growing personalized learning needs of students. In view of this, this article designs an intelligent translation skill training and evaluation system based on deep learning (DL), injecting new impetus into French language teaching using advanced technology. This system can provide diverse translation exercises based on students' specific abilities and needs, encouraging students to challenge themselves and make continuous progress in practice. At the same time, it can provide real-time feedback and evaluation of students' performance, helping them to grasp their own translation skills in real time, identify problems, and make improvements. Experimental results have shown that the system can effectively evaluate students' translation quality, not only improving their translation skills, but also enhancing their learning interest and enthusiasm.