| Summary: | 博士 === 中原大學 === 電子工程研究所 === 95 === The continual development of information technology in recent years has ensured its increasingly widespread use in many domains. Recent developments pertaining to the Internet and in computer technology have resulted in e-learning—a pedagogy that is free from time and space constraints. The issues of helping learners to study efficiently through lecturing procedures and using learning systems are now becoming increasingly important. Learning strategy based on a combination of the concept graphs learning system and cooperative learning is an important trend in computer and network aided instructions. The first step in cooperative leaning activities is to divide learners into groups. This dissertation proposes a grouping strategy to divide learning activities into several phases. The concept graph diagnostic system is adopted to evaluate learners’ learned concept nodes after each learning phase. The results of the evaluation are encoded to learning genes and used to calculate the group complementary score. A genetic algorithm is used to group the learners into learning groups according to the group complementary score.
On the basis of the evaluation results and prior experience, it can be stated that the grouping strategy along with the knowledge structure can result in more effective learning groups. However, the SPRT (sequential probability ratio test) has a drawback in that it cannot differentiate between the learners’ partial learning nodes. This leads to inaccuracies in the results of computing group complementary scores and may result in the creation of ineffective learning groups. In addition, apart from applying the knowledge structure to form learning groups, the interaction and learning efficiency are affected by other factors like roles in the group. Thus, this dissertation attempts to develop a partial learning evaluating strategy based on the degree of discrimination of the item bank. Thinking styles are also considered to be integrated into the grouping strategy that is based on concept graphs.
The heterogeneous grouping strategy based on learners’ knowledge structures and thinking styles is adopted to form learning groups. According to the evaluation results, this dissertation reveals that learners with the same level of learning achievement appear to have different levels of activeness and motivation toward e-learning. Learning is not merely a part of cognitive ability; the activeness and motivation of the learner also affect the learning results. Thus, this dissertation also proposes a learning portfolios diagnostics system whereby the learning log explorer can help teachers to efficiently understand and analyze learners’ online learning portfolios. In addition, the learning log explorer can gauge the online learning behaviors related to the learners’ learning achievements. The evaluation of confidence between the learning state and learning achievement yields positive experimental results. Further, the use of a supervisory agent enables teachers and learners to obtain their learning statuses or information provided by the proposed system via both the Web and email.
|
|---|
