摘要
专注是产生有效学习的先决条件,是取得良好学习成效的重要保证,在以自主学习为主的在线学习场景中具有更为重要的作用,但在线学习时空分离的特性难以保证学习者的专注度得到及时监控,故而探究精准识别在线学习专注度的可行方法至关重要。该研究主要关注在线学习中学习者的面部线索与眼动特征,基于从视频数据中提取的眼部视线、头部姿态、面部动作单元等面部线索特征以及从眼动数据中提取的注视停留时间、注视点、眼跳等眼动特征,分别通过两类单模态特征以及联合二者的多模态特征进行学习专注度识别,采用常用的六种机器学习方法构建相应的评估模型,对六种分类器的专注度预测性能进行了比较,并判断了专注度与学习成效的关系。实验结果表明,相较于面部线索,眼动特征具有更好的识别潜力,其体现的信息加工机制更能反映心理资源的投入程度;与单一模态相比,模态融合能显著提高学习专注度识别效果,揭示了面部线索特征和眼动特征对学习专注度识别的互补性;而学习专注度与学习成效显著相关,故而可将专注度作为优化在线学习的主要抓手,从学习材料设计者、教师与学习者等不同主体出发改善在线学习过程,提升在线学习效果。
Concentration is a prerequisite for effective learning and an important guarantee for achieving good learning results,which plays a more important role in the online learning scene dominated by independent learning.However,the space-time separation of online learning cannot ensure that learners’concentration can be monitored in time.Therefore,it is of great importance to explore feasible methods for accurate identification of online learning concentration.This study focused on learners’facial cues and eye movement features in online learning.Based on facial cues such as eye gaze,head posture and facial action units extracted from video data and eye movement features such as f dwell,gaze point and eye jump extracted from eye movement data,two kinds of single-mode features and the combination of the two multi-mode features were used to identify the learning concentration.Six kinds of commonly used machine learning methods were used to construct the corresponding evaluation model,and the prediction performance of the six kinds of classifier was compared.The relationship between the concentration and learning effectiveness was also judged.The experimental results showed that compared with facial cues,eye movement features have better recognition potential,that is,the information processing mechanism reflected by eye movement features can better reflect the investment degree of psychological resources.Compared with single mode,modal fusion can significantly improve the prediction effect of learning concentration level,revealing the complementarity of facial cues and eye movement features for learning attention recognition.In addition,learning concentration is significantly related to learning effectiveness,so concentration can be used as the main grip to optimize online learning,improve online learning process from different subjects such as learning material designers,teachers and learners,and enhance online learning effect.
作者
武法提
赖松
高姝睿
李鲁越
任伟祎
Wu Fati;Lai Song;Gao Shurui;Li Luyue;Ren Weiyi(School of Educational Technology,Faculty of Education,Beijing Normal University,Beijing 100875;Engineering Research Center of Digital Learning and Educational Public Service,Ministry of Education,Beijing Normal University,Beijing 100875)
出处
《中国电化教育》
CSSCI
北大核心
2022年第11期37-44,共8页
China Educational Technology
基金
国家自然科学基金面上项目“同步直播课堂中基于多模态数据的学习者专注度评估及其演化机制研究”(项目编号:62177008)研究成果。
关键词
学习专注度
眼动
面部线索
学习分析
多模态数据
learning concentration
eye movement
facial cues
learning analysis
multimodal data
