The aim of this study is to develop a method of qualitative analysis for extracting characteristics from e-Learning contents, in consideration of the semantic relationship among figures, sentences and sound-narrations. In order to extract the characteristics, we propose a method for contents construction. In general, we can expect higher degree of understanding of the frame contents by adding appropriate sound-narrations. In this research, we examine this effect under some experimental conditions. Especially, we explore the method for the content construction by adding sound-narrations for each frame in contents. We inserted the blended explanation by sound-narrations in each of the related frames, for the purpose of investigating synergy effect.

This research aimed to clarify the difference of learning effect according to learner’s property by distance learning using eye-contact type, conventional non-eye-contact type video conferencing systems and the real class by facing. The learner’s property was obtained beforehand, learning behavior factor and the learning effect was obtained from each class. As a learning model by simultaneous lecture, the recursive model was assumed, covariance structure analysis was executed. As a result, the following have been understood. At the real class, the learner with high “grounding” raises the learning effect. At the conventional type, the learner with high “maladaptive tendency” and “superiority complex” lower the learning effect. At the eye-contact type, the learner with high “extroversion” raises the learning effect, while the learner with high “insensitivity” lower the learning effect.

We developed and evaluated a system that manages student learning status by executing an agent on multiple lecture PCs to collect the progress status and results of simulated examinations included in electronic textbooks. The agent program constantly monitors the execution status of electronic textbooks on lecture PCs, detects the time when a student finishes the simulated test, and automatically captures the image data of the test result screen. Then, the agent performs template matching of the obtained image data, converts the image into score data, and registers the image and the data in the learning management database over the network. We ran the agent program on more than 800 exercise PCs and used them in a literacy course for 307 students at the university. Our results indicate that the group using our system showed significantly faster test score improvement and obtained a higher average score compared with the control group.

Beginners of a programming language sometimes lack syntactic knowledge to correct syntax errors in their programs. They repeat the same mistakes unless missing knowledge is given. We classified beginner’s syntax errors by supposed causes of the errors; knowledge shortage and carelessness, and implemented a support system for syntax errors.

In this paper, we present the support system that gives error messages including syntax knowledge to a learner of Prolog. This support system identifies the Prolog syntax errors by matching with 65 bug patterns. We evaluated the support system by comparing classes using this support and classes not using it. As a result, the syntax errors supported by error messages with syntax knowledge have decreased greatly in the supported classes compared with the unsupported classes. On the other hand, the syntax errors supported by messages with no knowledge have hardly decreased.