Information Sciences

Permanent URI for this collection

http://etd.aau.edu.et/handle/123456789/27

Browse

Browsing Information Sciences by Subject "Adaptive E-Learning Design"

Now showing 1 - 1 of 1
  • No Thumbnail Available
    Item
    Adaptive E-Learning Design By Artificial Neural Network Techniques: A Case of Ethiopian Higher Learning Institution
    (Addis Ababa University, 2010-06) Beyene, Melkamu; Lamnew, Workshet (PhD)
    An e-learning system is expected to recognize the different learner characteristics, the complex learning process that can be influenced by learner characteristics such as previous knowledge, learning styles, background, etc. The system is also expected to analyze students need to use learning material and an order of presentation that depends upon their own characteristics and needs. More adaptivity is expected from e-learning systems since students need to be considered independent. The purpose of this study is to investigate and demonstrate the possibility of developing Adaptive e-learning model for Ethiopian higher learning institutions learners. In this research, an experiment was made to build a learner model for an adaptive e-learning model. Firstly, Information independent of the course like learner goals and background and experiences as well as domain dependent data like prior knowledge about the course and learner expectations from the course were assessed. Secondly, an experimental learning styles predictive model is built through neural network data mining modeling technique. From the analysis of the assessment in this study, Ethiopian higher learning institutions learners’ students have varying learning goals, learner background and experiences, preferences, learning styles as well as prior knowledge about the course and their expectations from the course. The learning style predictive model is experimented by neural network data mining technique from student learning style questionnaire that received from 1296 respondents. Fourteen variables were selected from the twenty variables for each model. From the total dataset, 80%, 10% and 10% of the data is used for training, validating and testing purpose respectively. Neural network model is built which can correctly classify 93.07%, 92.30%, 96.15% and 94.61% of the validation set for visual-auditory, sequential-global, activist-reflective, sensitive-intuitive and learning styles respectively. Lastly, an adaptive e-learning model prototype was designed.