As for e-Learning, it is important to provide learning support tailored to each learner’s individual learning style. Normally, e-Learning support has been provided by humans through a process that involves identifying learners’ learning styles. However, it has been difficult for computers to replicate this process automatically.

Recently, generative AI has gained attention in the field of education. Ideally, generative AI could support learners by identifying their learning styles in the same way as humans do. However, relying entirely on AI for learning support is not effective. Nevertheless, generative AI can read learners’ learning data and extract their individual learning contexts as text. Considering that ability, we supposed that it would be possible to provide learning support similar to that provided by humans by feeding generative AI with learning styles predicted through machine learning trained on learning log data, learning contexts extracted by generative AI, and learning-support scenarios created by humans. In this study, we developed a generative-AI-based system that offers learning support tailored to learners’ individual learning styles and contexts of e-Learning.

The developed system consists of four functions:
- Learning-style-identification function:
A learner’s learning style is identified at the end of a course by using classification based on a supervised learning model applied to their learning log data.
- Learning-support-scenario-selection function:
An appropriate learning-support scenario is selected on the basis of the learner’s learning style and their current learning step.
- Navigation-and-facilitation-generation function:
Generative AI is provided with the necessary learning-related information to extract learning contexts based on the selected learning-support scenario and generate learning-support messages.
- Navigation-and-facilitation-execution function:
The generated messages, offering facilitation or guidance tailored to their learning style and learning steps, are presented to learners.

The system operates as follows. First, the learner is prompted to recall the content of the previous course, confirm the current learning goal, and set their own learning plan. At that point, the learning-support-scenario-selection function determines the appropriate support scenario. According to this scenario, the navigation-and-facilitation-generation function inputs LMS content data and the learner’s learning log data into the generative AI. The navigation-and-facilitation-execution function then offers facilitation to support the learner in planning their study. During e-learning, in accordance with the learning-support scenario, the learner's study plan and learning data are input into the generative AI. Guidance to help the learner revise their study plan and facilitate their learning is then provided. When the course is completed, the learner reflects on the course. At this point, in accordance with the learning-support scenario, the learning style identified by the learning-style-identification function, the learner's study plan, and learning data are input into the generative AI, which then generates and presents provides that encourage the learner’s reflection.

We evaluated whether the system could provide support tailored to learning styles and contexts by practical trial. The results of the trial confirmed that the system successfully provided learning support according to the learning-support scenario.

Keywords: e-learning, Learning style, Generative AI, Personalized learning support, Learning-support system.