Skip to main content Skip to navigation

FOKE: A Personalized and Explainable Education Framework Integrating Foundation Models, Knowledge Graphs, and Prompt Engineering

Project Overview

The document presents FOKE, an innovative framework that integrates large language models (LLMs) and knowledge graphs to transform educational practices by enhancing personalization, interactivity, and interpretability in learning experiences. FOKE tackles key challenges in education through its components, including hierarchical knowledge representation, multi-dimensional user profiling, and structured prompt engineering. It demonstrates its effectiveness in various applications such as programming education, personalized learning path planning, and intelligent writing assessment. By leveraging generative AI, FOKE aims to create more engaging and tailored educational experiences, ultimately revolutionizing how students learn and interact with educational content. The findings suggest that such AI-driven approaches can significantly improve educational outcomes by making learning more accessible and customized to individual needs.

Key Applications

Personalized Learning and Feedback Systems

Context: Providing interactive programming tutorials and recommending personalized course sequences based on learner profiles. This includes generating prompts for programming tasks and suggesting learning paths to help learners navigate their educational journey.

Implementation: Utilizes AI technologies to generate personalized prompts and learning paths by analyzing learner profiles and course information. This includes automating feedback on writing assignments and providing insights based on predefined criteria.

Outcomes: Enhances understanding of programming concepts, improves problem-solving skills, aids in navigating course selections, and promotes writing skills through actionable feedback.

Challenges: Requires continuous updates to knowledge bases, accurate user profiling, dynamic course information, and may struggle with subjective aspects of writing that necessitate human interpretation.

Implementation Barriers

Technical

The complexity of integrating multiple AI technologies into existing educational systems

Proposed Solutions: Iterative development and pilot testing to refine system capabilities

Data

Limited and domain-specific data may not capture the diversity of learners

Proposed Solutions: Incorporating a broader range of educational data for comprehensive profiling

Evaluation

Lack of standardized evaluation methods for comparing different educational AI systems

Proposed Solutions: Collaboration with domain experts to establish effective evaluation criteria

Project Team

Silan Hu

Researcher

Xiaoning Wang

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Silan Hu, Xiaoning Wang

Source Publication: View Original PaperLink opens in a new window

Project Contact: Dr. Jianhua Yang

LLM Model Version: gpt-4o-mini-2024-07-18

Analysis Provider: Openai

Let us know you agree to cookies