| المستخلص: |
With the spread of e-learning platforms across the Internet, including the large online learning courses called MOOCs (Massive Open Online Courses), there is a wide range of learning resources related to different educational topics and courses. Therefore, learners with different backgrounds, culture, skills, and knowledge level, are able to follow educational online courses to help them to acquire knowledge about different topics, develop their abilities and enable them to master different skills, without being limited to time or place. However, this openness characteristic of MOOCs can be a reason for overwhelming learnrs in the middle of mass learning resources. Another important aspect is related to the fact that the current offered courses are mainly developed based on teacher oriented approach. This means the contents and the assessements resources are mainly based on the course author’s experience. As such, the learning outcomes of each course are static and based on what have been introduced by the course author. This can be considered as shortcome as the course contents is not suitable to the learner’s knowledge or the course contents might not satisfy the expected or intended learning outcomes form the learner’s point of view. To overcome previous obstacles, there are various efforts and attempts have been proposed in the litreture to provide adaptation techniques to MOOCs. Proposed adaptation techniques are mainly related to dynamically adapt learning resources, assessment tools, content presentation, logical sequence of learning path using data mining and classification algoritms. Similarly, this thesis aims at providing an approach to support adaptation to content and assessment tools to cater learners’ needs. The proposed approach in this research work is innovatiove as it considers adaptation from twfold aspects. First, it utlizes the Support Vector Machine algorithm to autmaticlly map learning resources and Intended learning outcomes. Second, it utilizes Fuzzy Logic algortihm to determine the levels of assessment, such as questions and examinations at different levels: easy, medium and difficult and generate exams based on the current knowledge level of the learner. To validate the proposed approach, a proof of concept has been developed by utlizing the two algorithms (SVM and Fuzzy Logic). Moreover, a dataset related to learning resources for a specific course in Coursera MOOC platform was collected to be classified using SVM according to their relevance to a number of identified learning outcomes. Moreover, a number of questions and exams were collected from the same course. The developed prototype was provided with the learning contents and assessements for a specific course and it was evaluated. The results were promising proved accurate and satisfactory outcomes for both classifications of learning resources and assessment tools as the accuracy indicators were 71.5%.
|
