基于在线学习行为数据的人格特质识别研究

Online Learning Behavior based Personality Recognition

获取学习者个性特征是实现以学生为中心的精准化、个性化教学的重要前提,而学习行为是分析学习者个性特征的重要依据。本研究以参加奥鹏公共研修平台在线学习者为研究对象,基于人格特质生成学习行为偏好假设,探索利用机器学习分类算法实现在线学习行为的人格特质识别;同时基于文献构建人格特质类型与在线学习行为之间的映射关系,采用Rapid Miner数据挖掘工具探索决策树、朴素贝叶斯和支持向量机三种算法对五种人格特质的识别效果。结果发现:决策树算法对人格特质类型的识别准确率高于其他两种算法,对大五人格特质的综合识别效果最好;不同人格特质识别灵敏度不同,尽责性人格特质类型的识别灵敏度最高,神经质人格特质最低。

Learners’ personality is critical for student-centered and personalized teaching, and their learning behavior is essential for analyzing learners’ personality. Based on the hypothesis that personality generates learning behavior preference, this study explored personality recognition based on online learning behaviors, using machine learning classification algorithm. Firstly, Big Five Personality questionnaire was administrated to online learners from Open Edutainment. Secondly, the mapping relation between personality traits and online learning behaviors was constructed based on literature and verified by Spearman correlation analysis. Finally, Rapid Miner was used to explore the identification effects of Decision Tree, Naive Bayesian, and Support Vector Machine(SVM) algorithms on five personality traits. The study found that the recognition accuracy and effect of the Decision Tree algorithm on each personality type is higher than the other two algorithms. Moreover, the sensitivity of different personality recognition is found to be different: the conscientiousness is the highest, and the neuroticism is the lowest.