A Framework to identity parameters derived from interactions that can be used to identify emotions in an eLearning environment

Abstract

Learning and emotions have a strong correlation. A student in good emotional state that encourages learning helps to grasp new concepts and learn effectively, and vice versa. In eLearning, emotions of a student are not addressed as it is addressed in traditional class room. As a stepping stone to develop an emotion adaptive learning system, this project focuses on identifying the parameters derived from learner's interactions with the system that affect emotions. Student's interactions are captured while interacting with the e-Leaming system. These interactions are converted into quantified parameters. These parameters are collected for all students under observation, and then analyzed which contribute to explain the performance of students. The theory behind this used is the fact that learning encouraging emotions contribute to productive learning and better performance. The infrastructure to capture data is a web based system and it is used to administer the system as well as keep persistent data required for the analysis. Also it is used to report the results of analysis. The student's desktop system will record interactions with the system once the student logs into the system and starts learning a subject, until logged out. An application installed in student's desktop will be run in the background to capture interaction information that will be converted to parameter values. These parameter values become the input for the analysis that is performed in the server. Data analysis method used is Multiple Regression Analysis. Parameter values for each student are collected and recorded in the server and the analysis is performed. The analysis will report the interested parties about the relevant parameters that affect emotions, at the requested significance level and will show how the analysis should be carried out again to arrive at a better result. In conclusion, it can be stated that there exists a positive correlation between emotions and learning. Also it can be concluded that the framework developed is capable of obtaining values for parameters derived from keyboard, mouse and navigational patterns related interactions to test and identify emotions in an e-Learning environment. Also, after testing and evaluating the framework, the conclusion derived was that at 90% confidence level, none of the parameters used in the experiment can be used identify emotions.

Learning and emotions have a strong correlation. A student in good emotional state that encourages learning helps to grasp new concepts and learn effectively, and vice versa. In eLearning, emotions of a student are not addressed as it is addressed in traditional class room. As a stepping stone to develop an emotion adaptive learning system, this project focuses on identifying the parameters derived from learner's interactions with the system that affect emotions. Student's interactions are captured while interacting with the e-Leaming system. These interactions are converted into quantified parameters. These parameters are collected for all students under observation, and then analyzed which contribute to explain the performance of students. The theory behind this used is the fact that learning encouraging emotions contribute to productive learning and better performance. The infrastructure to capture data is a web based system and it is used to administer the system as well as keep persistent data required for the analysis. Also it is used to report the results of analysis. The student's desktop system will record interactions with the system once the student logs into the system and starts learning a subject, until logged out. An application installed in student's desktop will be run in the background to capture interaction information that will be converted to parameter values. These parameter values become the input for the analysis that is performed in the server. Data analysis method used is Multiple Regression Analysis. Parameter values for each student are collected and recorded in the server and the analysis is performed. The analysis will report the interested parties about the relevant parameters that affect emotions, at the requested significance level and will show how the analysis should be carried out again to arrive at a better result. In conclusion, it can be stated that there exists a positive correlation between emotions and learning. Also it can be concluded that the framework developed is capable of obtaining values for parameters derived from keyboard, mouse and navigational patterns related interactions to test and identify emotions in an e-Learning environment. Also, after testing and evaluating the framework, the conclusion derived was that at 90% confidence level, none of the parameters used in the experiment can be used identify emotions.