搜索
题目：问题验证标准对由Generated AI所产生的多选题之质量效果 --- 基于三门本科课程的初步实例探讨
报告人：王增相 阿萨巴斯卡大学管理学院
简介： 本项研究尝试了将问题验证标准（QVC）直接嵌入LLM提示中，以自动生成带反馈的多选题（MCQ），用于自适应形成式评估， 并且由专家评估 验证标准对所选的问题的质量之影响。问题生成应用程序建立在一种新的基于网络的App，应用于三门本科课程的MSQ制作。主题专家的验证显示，在提示中嵌入QVC和不嵌入QVC的问题之间，产生的质量没有统计学上的显著差异。这些发现有助于越来越多的研究表明，仅凭提示工程(Prompt Engineeing)可能不足以改进以提高课程教育质量为目的LLM。
报告人简介：Dr. Zengxiang Wang is full professor of finance at Athabasca University. His research touches on a range of topics, including shareholder activism, CEO compensation, director compensation, financial regulations, co-regulations, financial frauds, and distance education. He has published academic papers in Journal of Management, Managerial Finance, Journal of Management and Governance, Journal of Financial Crime, International Review of Financial Analysis, Distance Education in China, professional oriented papers in Ivey Business Journal and Canadian Investment Review, and business case studies in Ivey Publishing and Harvard Business Cases. He has also been involved in supervising five doctoral students. In addition to his teaching and research experience, Dr. Wang won faculty wide research award and professional association sponsored best paper award and education award and federal government community enhancement award. He was a trustee for 11.5 years supervising the operation and investment of a pension plan for faculty and staff of the research universities in a large Canadian province.
报告时间：2025年6月23日 15：30-17：30
报告地点：文渊楼 B434
主办单位：数学与统计学院
学院宣传片
