Framework for analyzing online asynchronous discussion by integrating content analysis and social network analysis

Abstract

Online Asynchronous Discussion (OAD) is a powerful way to conduct online conversation and a significant component of online learning. Unfortunately, existing Learning Management System (LMS) that generally provides online discussion cannot afford a comprehensive evaluation on the content of the transcripts and the level of interaction among participants. Therefore, this research explores the analysis process of OAD qualitatively and quantitatively. The work focuses on Content Analysis (CA) and Social Network Analysis (SNA), two popular methods employed by educators and researchers to analyze online discussion in e-learning environment. Although these two methods are well established, the techniques remain manual. Furthermore, presently, these two methods of analysis are conducted and studied independently. Hence, this research proposes a new framework integrating CA with SNA called CASNA, which provides comprehensive information of the result, and automation of the processes. CASNA is applied and embedded in LMS (Moodle) to validate the proposed framework. This research also introduces sentence as the unit of interaction instead of message to assess the level of participation among students. In addition, in order to qualitatively analyze the online discussion, two text classifiers; the Support Vector Machine (SVM) and the Back-propagation Neural Network (BPNN) approaches are employed to categorize the sentences based on Soller’s model and the results are compared. The evaluation of these two classifiers is done based on precision, accuracy, recall and F-Measure. The result shows that SVM outperform BPNN in terms of precision and accuracy; falls behind BPNN in terms of recall and F-Measure. This research also discusses the use of network indicators of SNA. Adjacency matrix, graph theory and network analysis techniques are applied to quantitatively define the network interactions among participants. This framework takes advantage of the strength of each method and offers dynamic analysis of the textual messages. It is expected to be more informative to educators as well as researchers in measuring the quality and quantity of OAD.