Scholarlink

It is undeniably true that student feedback is one of the most powerful components in learning as it allows teachers to understand to what extent their courses are assimilated. This scientific paper seeks to explore, analyze and enhance students ‘experience on MOOCs platforms using the BLOOM’s Taxonomy mainly through their feedback on MOOCS’ forums. BLOOM’s taxonomy is regarded as a great tool to build on cognitive knowledge. We have used synthetic datasets of feedback in order to train a language model to classify entries based on their types (question, answer, opinion) and then we followed through with classifying questions only with regard to the Bloom’s terminology. We aimed to generate feedback with their corresponding classification from Bloom’s classification using a large language model and fine tuning another one with few shot learning to perform this classification. Our paper’s contribution is expected to bring practical implications and recommendations for MOOCs’ platforms.

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

Breaking the MOOCs Barrier: A Bloom’s Taxonomy Driven Approach to Enhance Feedback Analysis with Few-Shot Learning and LLMs to Optimize Course Design

  • 章节

  • 发表日期：

    2025年

查看全文
  • Aicha Marrhich,
  • Ichrak Lafram,
  • Naoual Berbiche

摘要

It is undeniably true that student feedback is one of the most powerful components in learning as it allows teachers to understand to what extent their courses are assimilated. This scientific paper seeks to explore, analyze and enhance students ‘experience on MOOCs platforms using the BLOOM’s Taxonomy mainly through their feedback on MOOCS’ forums. BLOOM’s taxonomy is regarded as a great tool to build on cognitive knowledge. We have used synthetic datasets of feedback in order to train a language model to classify entries based on their types (question, answer, opinion) and then we followed through with classifying questions only with regard to the Bloom’s terminology. We aimed to generate feedback with their corresponding classification from Bloom’s classification using a large language model and fine tuning another one with few shot learning to perform this classification. Our paper’s contribution is expected to bring practical implications and recommendations for MOOCs’ platforms.