人工智能大模型技术在教育考试全题型阅卷中的应用  

Application of Artif icial Intelligence Large Language Modeling Technology in All-Question Marking of Education Exams

作　　者：王冠 魏兰 WANG Guan;WEI Lan

机构地区：[1]可之(宁波)人工智能科技有限公司,宁波315000 [2]上海交通大学宁波人工智能研究院 [3]布朗大学 [4]汇隽可之(深圳)人工智能科技有限公司

出　　处：《教育测量与评价》2024年第3期3-18,共16页Educational Measurement and Evaluation

摘　　要：随着数字技术的不断发展,教育考试阅卷工作正迈向智能化时代,AI大模型技术在阅卷中的应用探索已成为研究热点和发展重点。汇隽可之将其在认知智能生成式大语言模型领域研发的AI核心技术应用于考试阅卷工作,经过3年的实践与完善,该系统在AI评卷速度、辅助定标成效、AI评分质量等方面均得到有效验证。目前,AI大模型技术在教育考试全题型阅卷工作中已成功实现辅助评分功能,未来有望在各类考试中逐步替代人工,最终实现AI全自动评分。With the continuous development of digital technology,China's education assessment field is progressively adopting intelligent technologies,and the application of AI in grading has become a focal point for research and future developments.Huijun Kezhi has implemented its advanced AI technology,originally developed for cognitive intelligence generative large language models,in examination grading.After three years of implementation,the effectiveness of this technology has been proven in several areas:the processing speed of AI in grading,cluster analysis to support evaluation,and the overall quality of AI-based grading systems.Currently,AI technology is capable of assisting in the grading of all types of examination questions,and it is anticipated that AI will eventually fully automate scoring processes across various examination formats.

关 键 词：人工智能 考试阅卷 自然语言处理 全题型自动评卷 分析聚类 辅助质检 

分 类 号：G424.74[文化科学—课程与教学论]