The Study of Knowledge Structure Discovery ：A Case Study of biquitous Computing

• Main Authors: Yeh,Chih-Chung, 葉志中
• Other Authors: Chen,Tsung-Teng
• Format: Others
• Language: zh-TW
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/29288744405555342748

碩士 === 國立臺北大學 === 資訊管理研究所 === 95 === Due to fast rise of internet, human kind information broadcast enters a new epoch. In addition to considerably low entrance barrier in acquiring information, which is comparing to the past, thus, user of information aptly suffers the so called “Lost in information” syndrome. Consequently, how to swiftly and efficiently filter out accurate information, becomes another important subject. As for someone new to a specific domain in research study, how to focus onto current development status of intended research study field, and speedily filter out core scholastic research literatures for that field, become key elements for elevating scholastic research efficiency. Although this subject was dealt by past scholars mostly through traditional Citation analysis, which intended to overcome this phenomenon, nonetheless, it had little effect towards the overall connectivity among all literatures and the unearthing of the decisive literature. At the same time, it simply cannot process large volume of information. This research proposes two enhanced methods in regard to the above mentioned phenomenon respectively, which are “Pearson correlation coefficient” and “Laplacian Kernel” core computation method. The latter computes the times that scholastic citation cross-reference each other. Then, through path analysis, this can be presented via visualized web diagrams. And this study intends to use the prior mentioned processes and provide a standardized flow process in the end. And this process is to achieve a highly efficient way of comprehensively presenting specialized domain structures existed in the scholastic literatures. So, after focusing on this structured diagrams and further interpretation, it is discovered that “Laplacian Kernel” performs better than “Pearson’s correlation coefficient” in both literature classification and the expulsion of redundant linkages.