Heterogeneous Link Prediction Technique in Personalized E-Learning System using SVM Gopalakrishnan T.*, Dr. Sengottuvelan P.**, Dr. Bharathi A.***, Lokeshkumar R.**** *Department of Information Technology, Bannari Amman Institute of Technology, Erode, India **Associate Professor, Department of Computer Science, PG Extension Center, Periyar University, Dharmapuri, India ***Department of Information Technology, Bannari Amman Institute of Technology, Erode, India ****Department of Information Technology, Bannari Amman Institute of Technology, Erode, India Online published on 9 November, 2016. Abstract Personalized learning system is an online educational system; it is used to predict the learner's most prepared learning styles. Preferences of learning styles are found through way of linking different types of materials over the web. Deal of these website links, the link prediction is introduced; it is a problem technique in many online social networks, online E-learning centers, its forums and communities. In general, the link prediction techniques deal with homogeneous way of predicting the user's learning style in online social networks and educational systems. However, a new challenging problem in link prediction is to analyze heterogeneous learner's (e.g. male and female) based on their preference of learning styles in selecting the links. A day-by-day research works are being incorporated in the technology-enhanced between heterogeneous entity and their perception level. This research work mainly focus on examining the information of different learning styles and prediction of more preferred learning styles links by classification models through Weka tool. The experiment was examined using various data mining classification models, where Support vector machine (SVM) classification model has achieved good performance result and has predicted the learner's potential by analyzing the most representative learner's’ features of perception level and gender. Finally the research work has concluded how these representative learner's features are incorporated with learning style dimensions as well as better learning model is used on the system. Top Keywords Link Prediction, Learning features, Heterogeneous, Data mining models. Top |