| Summary: | 碩士 === 中華大學 === 資訊管理學系(所) === 96 === E-learning has become a new trend as the Internet technology progresses. It has been investigated that learners’ individual characteristics would affect their learning outcome when they learn from learning websites, and they would achieve better learning outcome if they are provided matched learning style. Due to this result, adaptive learning system becomes a popular research topic these years. The most important question in developing adaptive learning system is how to construct a valid learner model. At present, there are many adaptive systems make the learners answer questions or fill in questionnaires first to confirm their cognitive style and then to achieve the adaptive outcome. However, this way would cause the inconvenience and trouble for learners. The system would memorize the information that the learner had browsed and provide adaptive web-sites when the learner get in next time; this would invade one’s privacy. Based on the problem, the study did not ask learners to answer any questions nor fill in any questionnaires, but only collected the browsers’ behavior to analyze their learning style via the learner model constructed by the neural network. The colleted browser behaviors included the click times, click proportion, and average retention periods of information presentation objects as well as network linked types. No privacy problems would exist for the reason that the browse information disappeared after the browser was closed.
The study would be presented in two parts. In the first part, learners’ cognitive style questionnaire and the information presentation objects and linked types of browser behaviors were collected. By using the classification function of neural network, a learner model was constructed to analyze learners’ cognitive styles through browsers’ behaviors immediately. Learners’ cognitive styles were then analyzed through the information objects and network linked types of browsers’ behaviors. There were four information presentation modes classified according to the learners’ different cognitive styles: interpersonal, mastery, self-expressive, and understanding styles. The conspicuous colors were used to display the information presentation objects that learners prefer. For the materials, if learners prefer to browse pictures, the adaptive websites would show pictures directly without learners’ clicks. In the second part, three groups of students were involved in the experiment. They browsed adaptive webpage with cognitive style on-line instant analysis, adaptive webpage with a known cognitive style, and non-adaptive webpage respectively and offered the influences on them, and the validity of learners model constructed by neural network were further analyzed.
Several conclusions of the research are as the following:
1. Learners of different cognitive styles have different preferred behaviors to browse the learning websites.
2. The learner model constructed by the neural network can effectively classify cognitive styles, and the average validity was 80 percent. Analyzing learners cognitive styles correctly would be helpful to the adaptive function in the learning websites.
3. It has great influences on learners when they browse adaptive webpage with cognitive style on-line instant analysis, adaptive webpage with a known cognitive style, and non-adaptive webpage.
|
|---|
