Hello everyone, I am Chu, a research student. In this article, I would like to introduce a paper I read in our most recent English Literature Seminar and share my thoughts on it.

• Paper Title: Taking adaptive learning in educational settings to the next level: leveraging natural language processing for improved personalization

• Journal: Educational technology research and development

• Year of Publication: 2024

• Authors: Mathias Mejeh & Martin Rehm

Self-Regulated Learning (SRL) is an extremely complex process, and providing personalized learning for each student is also highly complex. How can we “connect the dots” between these two intricate processes and better realize theory through technology? This paper provides several insights by utilizing Natural Language Processing (NLP) technology.

To support “Self-Regulated Learning”—the ability for students to manage their own learning—”Adaptive Learning Technology (ALT)” is drawing attention for its capability to provide learning methods tailored to individual students. Using ALT, it is possible to provide materials and learning strategies suited to each student and show a path to achieving goals. However, effectively utilizing ALT to support student learning is not easy, and many challenges remain. In this study, we explored methods to combine ALT with NLP to provide real-time support based on individual learning situations and improve student learning.

What is Self-Regulated Learning (SRL)? SRL is the process by which students plan, execute, and adjust their learning while reflecting on the results. For example, thinking about “which subject to study and for how long” or checking “progress” while actually studying are parts of SRL. These must be performed proactively by the learner, and when schools and educational technology support this, learning becomes more effective.

What is Adaptive Learning Technology (ALT)? ALT is a technology that utilizes Artificial Intelligence (AI) and machine learning to grasp students’ progress and weak points, providing appropriate materials and advice. For example, it provides suitable learning methods for each individual, such as giving difficult problems to students who excel in mathematics and providing more basic problems to those who struggle. Additionally, it creates an environment where students can learn with confidence by visualizing progress and showing steps to achieve goals.

Strengthening ALT with Natural Language Processing (NLP) NLP is a technology that allows computers to understand and process text and language. In this study, by incorporating NLP into ALT, we realized real-time support tailored to the student. For instance, by analyzing comments or answers written by students online, it is possible to identify “current sticking points” or “content to be learned next” and provide individual advice. Furthermore, by analyzing student emotions, it is possible to respond to changes such as “motivation for learning is decreasing.”

How was it actually conducted? This study conducted an experiment with 69 Swiss high school students using a digital tool called “studybuddy.” In “studybuddy,” data regarding the learner’s motivation, emotions, cognition, metacognition, and resource management are collected through surveys using a 7-point Likert scale to support SRL. The surveys are integrated into the digital learning environment, short daily surveys are conducted, and individual adjustment strategies are provided based on the students’ self-efficacy responses. This tool proposes advice and plans to students based on their learning status and features functions such as:

• Automated Prompt System: Sends helpful advice and notifications during learning.

• Learning Dashboard: Displays student progress in an easy-to-understand manner.

• Personalized Strategies: Proposes approaches tailored to each student’s learning style and needs.

• Planning Tool: Supports learning plans with calendar and memo functions.

By using these functions, students found it easier to grasp their own learning status and were able to learn efficiently.

The system integrates with Learning Analytics (LA) to provide appropriate individual feedback for each learner. These strategies cover support regarding motivation, emotional regulation, cognition, and task management, and are customized using the German version of the Learning and Study Strategies Inventory (LASSI). The automated feedback system and the format of strategy provision aim to help learners reflect on and adjust their learning process. The co-design process focused on adaptability, strategy communication, and improving the user interface, aiming to meet the individual needs of learners.

Research Results In the initial stages of the research, we analyzed data to see how students were performing self-regulated learning. As a result, it was found that learning methods and progress varied greatly among students, confirming that personalized support is possible by utilizing ALT. Additionally, through analysis using NLP, we were able to grasp the learning state at that time from the words and emotional expressions used by students during learning. For example, when words like “difficult” or “don’t understand” were used frequently, it became possible to take specific actions, such as proposing more basic materials to that student.

My Thoughts The methods in this paper are very helpful as references, especially the techniques such as opinion mining, part-of-speech tagging, and sentiment analysis used to analyze metacognition in individually optimized software, which provide very interesting approaches. However, since this paper is limited to field needs analysis and lacks data such as user behavior logs, I felt the quantitative analysis part was somewhat weak.

There are several challenges in this study. At this point, the automated prompt system, digital dashboard, and personalized strategies/planning tools have not yet been included. In particular, it is not yet clear what role the automated prompt system plays in Self-Regulated Learning (SRL). Furthermore, while preceding research mentions natural language learning, it does not clarify exactly what functions or software the designed software uses through natural language learning. I thought this part was for analyzing interview questions, but it seems that is not actually the case.