Kwame for Science:基于Sentence-BERT的西亚科学教育AI助教
计算与语言
2022-07-12 v2 计算机与社会
人机交互
摘要
非洲师生比极高,限制了学生接触教师的机会,导致学生难以获得问题解答。本工作中,我们扩展了先前的AI助教Kwame,将其适配于科学教育,并部署为网页应用。Kwame for Science基于西非高级中学证书考试(WASSCE)的综合科学科目回答学生问题。它是一个基于Sentence-BERT的问答网页应用,针对科学问题显示3段作为答案的文字及置信度分数,此外还展示前5个相关的历年考题及其答案。我们对Kwame for Science进行了为期2.5周的真实部署初步评估,显示前3准确率为87.5%(n=56),覆盖11个国家的190名用户。Kwame for Science将能为非洲数百万民众提供可扩展、低成本且优质的远程教育。
引用
@article{arxiv.2206.13703,
title = {Kwame for Science: An AI Teaching Assistant Based on Sentence-BERT for Science Education in West Africa},
author = {George Boateng and Samuel John and Andrew Glago and Samuel Boateng and Victor Kumbol},
journal= {arXiv preprint arXiv:2206.13703},
year = {2022}
}
备注
5 pages, Accepted at the Fourth Workshop on Intelligent Textbooks (iTextbooks) at the 23th International Conference on Artificial Intelligence in Education (AIED 2022)