Scholarlink

This paper introduces an approach to evaluating learning path recommendations in e-learning systems. It employs the capabilities of AI-generated personas by large language models (LLMs) to assess learning paths that align with user preferences and behavior. The proposed system uses Wikipedia clickstream data to recommend learning paths through Wikipedia articles. An LLM, such as GPT-4, simulates expert assessments to validate the learning paths, addressing scalability and efficiency challenges in statistical data collection from human experts for validation. This integration reduces reliance on human expertise and enables real-time evaluation. The system architecture uses MongoDB for tree-like data storage and management, and Flask for streamlined automation of the validation process.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Automated Learning Path Evaluation with AI-Generated Judges

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Parvathy Menon,
  • Nuha Aburamadan,
  • Ahmed Ali Seyam,
  • Arash Kermani Kolankeh

摘要

This paper introduces an approach to evaluating learning path recommendations in e-learning systems. It employs the capabilities of AI-generated personas by large language models (LLMs) to assess learning paths that align with user preferences and behavior. The proposed system uses Wikipedia clickstream data to recommend learning paths through Wikipedia articles. An LLM, such as GPT-4, simulates expert assessments to validate the learning paths, addressing scalability and efficiency challenges in statistical data collection from human experts for validation. This integration reduces reliance on human expertise and enables real-time evaluation. The system architecture uses MongoDB for tree-like data storage and management, and Flask for streamlined automation of the validation process.