语言智能场景下的学习行为多模态情感语义库建设  

Building a Multimodal Emotional Semantic Database for Learning Behavior in Language Intelligence Scenarios

作　　者：周楠 周建设[1] ZHOU Nan;ZHOU Jianshe

机构地区：[1]首都师范大学文学院/中国语言智能研究中心,北京100048 [2]北京开放大学首都终身教育研究基地,北京100081

出　　处：《语言文字应用》2023年第3期121-131,共11页Applied Linguistics

基　　金：国家科技部科技创新重大项目“复杂版面手写图文识别及理解关键技术研究”(2020AAA0109700);国家社科基金项目“明代女性作家考录”(19BZW074);国家社科基金项目“基于语言智能的国际中文测试话题——词汇图谱研究”(22CYY040);中国教育技术协会重大项目“中文表达能力(CEA)标准研制及其智能测评应用创新研究”(XJJ202205003)的支持。

摘　　要：学生学习行为多模态情感语义库建设对学生学习情感分析与教学评价具有重要意义,它可以通过文本、语音和视频等多模态信息分析和识别学生情感,有助于掌握学生的学习状态。目前,学习行为情感语义库较少且形态单一,制约了学习行为情感分析技术的进一步发展。本研究基于语言智能场景制定了一套科学完整的标注体系,构建了学习行为多模态情感语义库智能建设流程,并探讨了语义库在解析情感表达和构造情感智能识别模型、开展教学评价等多个方面的应用,为提高语义库的构建效率、掌握学生学习状态、优化教学策略提供了支持。The construction of a multimodal emotional semantic database for student learning behaviors is of great significance to analyzing and evaluating students'learning emotions and teaching evaluations.It can analyze and recognize students'emotions throughmulti-modal information such as text,voice,and video,thus helping to grasp the learning status of students.Currently,there are relatively few and uniform forms of learning behavior emotional semantic databases,which restrict the further development of learning behavior emotion analysis technology.This study developed a scientific and comprehensive annotation system based on language intelligence scenarios,constructed an intelligent construction process of the multimodal emotional semantic database for learning behaviors,and explored the applications of the semantic database in parsing emotional expressions,constructing emotional intelligence recognition models,conducting teaching evaluations,and other aspects.This provides supports for improving the efficiency of constructing semantic databases,mastering students'learning status,and optimizing teaching strategies.

关 键 词：语言智能 学生学习行为 情感语义库 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]