Distribution of Educational Materials and Digital Twins, OKLM Development, Recommendation Functions, and Group Formation Tools

This research project aims to develop a knowledge model and learner model that enable consistent evaluation and comparison of learning content across the diverse educational materials and learning support systems that learners engage with throughout their lives.

By extracting knowledge elements from educational materials and managing and linking them in a unified way, this project enables the integration of learning data across different learning support systems.

In addition, by using OKLM to estimate learning history and proficiency levels, we aim to realize a new form of learning support that recommends optimal learning materials and suitable learning partners to each learner.

Through these efforts, we aim to provide personalized and optimal learning support, as well as a transparent and fair educational environment.

About OKLM

The Open Knowledge and Learner Model (OKLM) can be described as a digital twin (DT) that reflects a learner’s knowledge state and learning-related characteristics.

OKLM creates models that estimate who has acquired what knowledge and through what kinds of learning behaviors, by linking a pre-constructed knowledge model—based on learning materials—with learners’ activity logs collected from learning support systems.

In this framework, learners and teachers can refer to the learner model and modify it as needed or under certain conditions—reflecting the model’s openness to learners.

Moreover, as long as a knowledge model can be generated from the learning materials, OKLM can handle knowledge from any domain—demonstrating the openness of the knowledge model.

Furthermore, by providing an API to external learning support systems, OKLM enables advanced learning support based on its internal data, as well as real-time updates to that data in response to learner behavior within those external systems—demonstrating its openness to external platforms.

OKLM is a learner model that incorporates all three forms of openness described above, while adding a knowledge-based perspective to the existing context of learning analytics (LA).

Example of an OKLM Digital Twin (DT): Knowledge Model (from learning materials) + Learning Logs (from a Learning Analytics system)

The knowledge model in OKLM is represented as a graph structure consisting of knowledge elements and the relationships between them.

The learner model is represented using estimated proficiency levels, which are calculated by adding information about each learner’s access to the corresponding knowledge elements in the knowledge model.

The generated learner model can be represented and visualized through the interface of the learning analytics system.

OKLM is a system designed to support learners in effectively constructing and applying knowledge, and it is built upon a knowledge model containing a vast number of knowledge elements and their interconnections.

This system has been applied to tasks such as research paper recommendation and English learning support, and has shown positive results in actual classroom settings.

In particular, demonstration experiments in English reading classes have shown that the system performs well as a tool for deepening learner understanding, and API development for external users is also underway.

We are also working on a new approach that analyzes learner characteristics based on the forgetting curve.

Use Case of OKLM Digital Twin technology: Forming Student Groups

The group formation system adopts an innovative approach to grouping students either homogeneously or heterogeneously based on various characteristics.

This system utilizes a mixed genetic algorithm to enable optimal group formation, with the aim of maximizing learning outcomes.

In active reading, the OKLM word cloud feature is used to visualize each group member’s linguistic knowledge and to facilitate group discussions.

Use Case of OKLM Digital Twin: Peer Support

The peer recommendation system promotes mutual support among learners by leveraging OKLM data to recommend optimal peer helpers. This system has two key features that encourage peer-to-peer support.

• Learners who have received help are more likely to be recommended to help others, thereby expanding the circle of support.
• The OKLM data used for recommendations is visualized in the user interface, allowing learners to refer to it when selecting helpers and thereby receive more appropriate support.

Return to Projects Page