| Summary: | 碩士 === 國立臺南大學 === 資訊工程學系碩士班 === 105 === Owning to the thriving development in artificial intelligence, digital self-learning platforms like Coursera and Khan Academy have become one of the learning resources for people to learn. This thesis uses Boyo 1200-word English learning materials as the source of the items on the “2015 Formosa Experience Activity” and “Formosa Cup 2015.” Additionally, this thesis also constructs the ontology of the Boyo 1200 English words to make learners easier to learn. We first use the Gauss–Seidel method to estimate the items’ parameters according to learners’ response pattern of the “2015 Formosa Experience Activity.” Then, learners pass through a series of level games to enhance the learning entertainment based on the adaptive test. Learners’ response pattern is the source of our experiments. The knowledge base and rule base are described using fuzzy markup language (FML) to infer the possibility of the correctly answering the item and further to recommend the suitable learning materials for them. Finally, machine learning is implemented to optimize the experimental results based on genetic algorithm (GA) and particle swarm optimization (PSO). At the same time, the convolutional operation is used to find the features of learners’ response pattern. From the experimental results, it shows that (1) PSO performs better than GA under the same evolving generations, and (2) the adopted items on “Formosa Cup 2015” are much easier for most of the learners after convolutional operation.
|
|---|
