Research On Emotion Cognitive Modeling Technology Based On Improved SVM In E - Learning

The research of this paper is supported by the National Natural Science Foundation of China(Emotional cognitive analysis and research based on studies expression in E-Learning,No.60970052)and Beijing Natural Science Foundation(The Study of Personalized e-learning Community Education based on Emotional Psychology,No.4112014)The main purpose of this paper is to solve the problem of emotion deficiency in traditional network teaching, to provide the analysis of academic emotions and the modeling of academic emotion recognition in emotional and cognitive interaction E-learning system.Analyze and model the facial features through collecting and extracting these features from the project group’s students under E-learning circumstances, and determine the learner’s emotional reactions generated by the current learning content.Then we get the learner’s academic emotional state, and provide emotional function of individualized emotional teaching E-learning system.The main work and achievements of this paper are as follows:Firstly,we investigate and analyze the learner’s academic emotions under E-learning circumstances, combining the basic theory of psychology and educational psychology, and further analyzed the relevant learning characters of learners under normal learning state in E-learning system.Based on these characters and influential factors, we build a academic cognitive model of emotion recognition from the aversion degree, focus degree, and pleasure degree combining OCC emotion model.Secondly, Every learners’ learning habits are different from others because of their unique personality traits which makes it difficult to recognize emotions. In order to improve the efficiency and accuracy of academic emotion recognition under special circumstances, the solution is to build a individualized emotional model for different learners.Experimental results show that the individualized emotional model can make judgment directly when learners under special learning conditions(such as eyes closed, etc.),which improves the efficiency of academic emotion recognition. We compare and analyze the value of four different facial expression when the learners under normal learning status. The analysis results can ensure the stability and availability of data which provides the basis for emotional cognitive models. Thirdly, Based on the OCC emotion model, using the nonlinear characteristics of small sample in network of the Support Vector Machine, we build a emotional cognitive model under E-learning circumstances and generate a comprehensive emotion in three-dimensional emotional space from the aversion degree, focus degree, and pleasure degree.In order to optimize the parameters of support vector machines, we use particle swarm optimization(PSO) to improve the SVM which resulting in more accurate recognition rate of academic emotions. The comprehensive emotions generated by the model provides important basis of reasoning teaching strategies in E-learning system for other students in project group.Finally, Using JAVA programming language in the Eclipse platform, we achieve a emotional cognitive prototype system for analyzing the learner’s academic emotions in E-learning combining the individualized emotional model and the comprehensive emotional model and get good experimental results.