Browsing by Author "Liu, Yi-Li"

Now showing 1 - 3 of 3

A Rule-based Learning Diagnosis Scheme to Scaffold Self-Regulated Learning for Hypermedia-based Learning
(2011)
Lin, Huan-Yu
;
Su, Jun-Ming
;
Tseng, Shian-Shyong
;
Liu, Yi-Li
;
資訊工程學系
Hypermedia-based learning (HBL) has shown the potential capabilities to improve learners' learning performance in terms of complicated subjects and skills, but most of learners are difficult to learn well in this kind of learning environment due to the lack of self-regulated learning abilities. The manually adaptive scaffoldings, which are the most effective approach to facilitate learners to regulate their learning strategies, are quite time and work consuming for teachers, and most of existing intelligent tutoring systems without modeling diagnosis knowledge of teachers are not able to provide learners with the adaptive scaffolding concerning learning barriers and corresponding reasons. Therefore, in this paper, a novel Rule-based Learning Diagnosis Scheme (RLDS) is proposed to automatically support the adaptive scaffoldings for learners in HBL environments, where rule base and concept ontology based upon teachers' diagnosis knowledge are developed to represent the subject concepts in order to analyze learning barriers. The experimental results showed that learners who studied with adaptive scaffoldings of RLDS had significantly higher scores in post-test than those who studied without adaptive scaffoldings.
OSCAR: an Online SCalable Adaptive Recommender for Improving the Recommendation Effectiveness of Entertainment Video Webshop
(2010)
Lin, Huan-Yu
;
Su, Jun-Ming
;
Liu, Yi-Li
;
Li, Jin-Long
;
Tseng, Shian-Shyong
A recommender system is beneficial for the sales of e-commerce, so many kinds of recommendation approaches have been proposed for various situations. However, each recommendation approach can deal well with some kinds of categories and users\' behaviors only. Accordingly, how to provide users with the personalized recommendation with higher fidelity is an important issue. Therefore, in this paper, an Online SCalable Adaptive Recommendation scheme, called OSCAR, has been proposed in order to take advantages of various recommendation approaches and then efficiently coordinate them to adaptively meet the users\' preferences according to the various contents\' characteristics and users\' behaviors. Besides, the experimental results show that OSCAR\'s recommendation effectiveness is better and more stable than existing approaches.
Program Peer Assessment Using Tagging Approach
(2008)
Liu, Yi-Li
;
Tseng, Shian-Shyong
;
Lin, Huan-Yu
;
Weng, Jui-Feng
;
Chu, Yian-Shu
Most of die programming novices feel hard to learn programming skills, such as coding, tracing, debugging and testing. To help them learn programming well, computer program assessments which can find out the novices' programming misconception have been proposed. Although traditional computer program assessment can scow die student's program by executing die program with test cases, it is ineffective because die novices' programming misconception in the coding mistakes and tracing mistakes can not be singled out In die peer assessment activity, die learners can discuss their works with each other, so every learner can get die comments; about his/her work together with die score. This paper proposes a new computer program peer assessment to help die novices easily find out their programming misconception in the coding mistakes and tracing mistakes. An experiment has been done to evaluate the effectiveness of die new assessment. We may conclude that die process is useful to single out die leamer's coding and tracing problems, and that the tracing guidance is useful to direct the novice to trace peer's program.