Skip to main content Skip to navigation

LLM-Driven Personalized Answer Generation and Evaluation

Project Overview

The document explores the application of generative AI, specifically Large Language Models (LLMs), in enhancing educational experiences, particularly in language learning and programming. It highlights the potential of AI to generate personalized responses in online learning environments, which can significantly boost learner engagement and alleviate the workload of educators. Through its evaluation of different strategies for generating these personalized answers, the study finds that incorporating user-provided examples of preferred responses markedly enhances the quality of the AI-generated content. This approach not only tailors educational materials to individual learner needs but also fosters a more interactive and effective learning atmosphere, underscoring the transformative role of generative AI in modern education.

Key Applications

LLM-driven personalized answer generation

Context: Online learning platforms, targeting language learners and programming students.

Implementation: Utilized StackExchange data to generate personalized answers via different prompting strategies including 0-shot, 1-shot, and few-shot scenarios.

Outcomes: Improved alignment of generated answers with user preferences, demonstrated by higher BERTScores and favorable evaluations from human experts.

Challenges: Limitations in handling real-world questions and the need for more effective personalization methods.

Implementation Barriers

Technical Barrier

Limitations in Large Language Models' ability to handle complex real-world queries.

Proposed Solutions: Further research and development of customized answering methods tailored to personalized education.

Implementation Barrier

The challenge of ensuring the quality and relevance of generated answers.

Proposed Solutions: Conducting evaluations using BERTScore and human assessments to validate the effectiveness of the generated responses.

Project Team

Mohammadreza Molavi

Researcher

Mohammadreza Tavakoli

Researcher

Mohammad Moein

Researcher

Abdolali Faraji

Researcher

Gábor Kismihók

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Mohammadreza Molavi, Mohammadreza Tavakoli, Mohammad Moein, Abdolali Faraji, Gábor Kismihók

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