Supporting Student Decisions on Learning Recommendations: An LLM-Based Chatbot with Knowledge Graph Contextualization for Conversational Explainability and Mentoring
Project Overview
The document explores the use of a generative AI-based chatbot in education to improve the explainability of learning recommendations, facilitating students' understanding of the rationale behind these suggestions through interactive conversations. By integrating a knowledge graph, the chatbot contextualizes its responses and provides tailored support, even connecting students with human mentors when necessary. While the implementation of this technology presents promising benefits for personalized learning experiences, the document also addresses significant challenges, including issues of accuracy, bias, and the necessity for transparency in the use of generative AI within educational settings. Overall, the findings underscore the potential of generative AI to enhance educational outcomes while also pointing out the need for careful consideration of its limitations to ensure effective and equitable learning environments.
Key Applications
LLM-Based Chatbot for Learning Recommendation Explainability
Context: Online learning environments, targeting students seeking guidance on learning paths.
Implementation: Developed a chatbot module integrated into a web application that uses LLMs and KGs to provide explanations for learning recommendations, facilitating interactive conversations with students.
Outcomes: Improved understanding of learning recommendations among students, increased user satisfaction with the chatbot's performance, and established a proof-of-concept for conversational explainability.
Challenges: Limitations of LLMs in generating accurate and relevant educational content, difficulties in contextualizing user queries, and the need for ongoing research to evaluate long-term effects.
Implementation Barriers
Technical Limitations
LLMs present challenges such as hallucinations, bias, and difficulty in generating contextually relevant responses. These technical issues can hinder the effective use of generative AI in educational contexts.
Proposed Solutions: Implementing model fine-tuning methods, contextualizing prompts using knowledge graphs, and developing robust evaluation frameworks to enhance accuracy and reliability.
Ethical Concerns
Lack of transparency in LLM outputs raises ethical issues, as outputs may not meet the diverse needs of educational stakeholders. This can lead to mistrust among educators and students.
Proposed Solutions: Involving educators and domain experts in the design and evaluation phases, ensuring that the outputs are pedagogically sound and aligned with educational standards.
Project Team
Hasan Abu-Rasheed
Researcher
Mohamad Hussam Abdulsalam
Researcher
Christian Weber
Researcher
Madjid Fathi
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Hasan Abu-Rasheed, Mohamad Hussam Abdulsalam, Christian Weber, Madjid Fathi
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