الفهرس 
chapter 1Only 14 pages are availabe for public view
 |  | from | 148 |  |  |
| | | from | 148 | | |
Abstract
There is an inherent social network in any E-learning system. Main actors in E-learning social networks are: instructors, learners, and researchers. Through the learning process, E-learning entities communicate in a social way with each other. Different sorts of communication can take place examples are many including: exchange of data, opinions, open discussions, forming groups for projects, assignments, and labs, etc. for decades, mining has been focused on the learning materials or on the student’s learning paths trying to find out more about student’s behavior. Emergence of social networks lately has reflected the way we interact in reality to the web. Mining social networks activities help to share interest between groups of people with common features.
In this thesis, we applied social networks analysis techniques over social networks formed in the E-learning process. In E-learning students interact with students, students interact with instructors, and instructors interact with instructors plus, if we consider the community of researchers to be part of the E-learning process, this adds a new community. Social network analysis leads to understanding many of the aspects in order to achieve cooperative and useful community at any field.
Recommender systems are needed to find subject items of one’s interest. We review recommender systems and recommendation methods. We propose a subject personalization framework based on adaptive hypermedia for computer science ACM curricula. We extend Hermes framework with subject recommendation functionality. We combine TF-IDF term extraction method with cosine similarity measure. specialization and standard subject database are incorporated into the knowledgebase. Based on the performed evaluation, we conclude that semantic recommender systems in general outperform traditional recommenders systems with respect to accuracy, precision, and recall and the the proposed recommender has a better f-measure than existing semantic recommenders.