There are various fortunes to improve E-Learning web applications regard to continuing challenges like their lack of interaction between teachers and students and structural evaluation of the presented learning activity. In this paper we gather E-Learning system web server log data and some information about its control panel first, and then we preprocess gathered data. After extracting association rules form such data and selecting results with more confidence coefficient we use them as virtual consultant for E-Learning system improvement. Our database for storing and extracting patterns is MySQL and extracting association rules with more confidence coefficient has been done by Weka Software. Selected E-Learning web application is Moodle and our virtual university case study is Iran University of Science and Technology. Results show proposes which is not accessible via human consultant and cannot be inferred from current application log directly.
