The development of a web browsing preference model based on rank order statistics using data mining tools

• Main Authors: Shu-Ting Weng, 翁淑婷
• Other Authors: Kuo-Hao Tang
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/21378378501541083455

碩士 === 逢甲大學 === 工業工程學所 === 92 === Overwhelmed by the amount of information on Internet, if a computer can judge the importance or relevance for a webpage being browsed, an intelligent agent program can then retrieve meaningful information accordingly. From human-computer interaction perspective, a user will use peripheral devices such as a mouse and keyboard to interact with a computer during browsing and from which browsing behaviors can be collected. This thesis is to develop a program running at the background using API calls and collect all interaction patterns under different browsing purposes. There are two stages in this research. The first stage is to analyze users’ mouse movements. The results show that user, direction, and distance are statistically significant, and mouse movement speed and acceleration can be used as a predictive model for distinguishing between users. A data mining tool was used to analyze the interaction patterns collected from web browsing tasks. The extracted user patterns showed that users slowed down the speed and moved backward and forth when moved into target area of a webpage. Using accuracy and coverage as performance indexes for these patterns, it can be shown that the accuracy and coverage were 41.90% and 53.52% respectively for positive detection, and they were 58.10% and 74.22% for correct reject.