Automated analysis of confidence, interests, and attentiveness in a computer assisted learning environment

• Main Authors: Ko-Yu Lin, 林科宇
• Other Authors: Yen-Teh Hsia
• Format: Others
• Language: zh-TW
• Published: 2003
• Online Access: http://ndltd.ncl.edu.tw/handle/ck8k7e

碩士 === 中原大學 === 資訊工程研究所 === 91 === Due to the rapid development of computer and hypermedia technology, it becomes all the more apparent that when used properly, we can develop good learning technologies. But no matter how we change our methodology of teaching and learning, there are certain characteristics of human learning that remain unchanged. How can we detect and evaluate the intellectual state of a learner? This may be a problem that will need to be dealt with by all hypermedia-based learning environments in the future. Some of the external behaviors of a learner may result from his/her internal intellectual state such as confidence, interests, and attentiveness. There is no way we can directly measure the degrees of confidence, interests, or attentiveness of a learner. But we can surely observe the external behaviors of the learner and try to diagnose his/her internal intellectual state. In constructing our hypermedia-based learning environment, we provided a variety of user interface. By tracking and sometimes inducing the learner’s system-usage behavior behind the scenes, we obtain a large quantity of usage profile. And then, by filtering out useful information and also be evaluating the information, we obtain meaningful estimates of the learner’s intellectual state as related to confidence, interests, and attentiveness. Though much has been done in the area of user-behavior tracking, we did not find much research in tracking and evaluating the learner’s confidence. Most of the research works that we found use domain-dependent methods for evaluation. In this thesis, we propose a domain independent method for evaluating the learner’s confidence, interest, and attentiveness. We also did a small prototype experiment. The results suggest that our methods are reasonably accurate.