Correlation between Customer Preference and Browsing Trace — A Field Study Using Cellular Phone Service as Research Vehicle

• Main Authors: Chang, Shu- Mu, 張樹木
• Other Authors: Chen, Tsung-Teng
• Format: Others
• Language: zh-TW
• Online Access: http://ndltd.ncl.edu.tw/handle/53969860138779579424

碩士 === 國立中正大學 === 資訊管理學系 === 87 === It has been postulated that the browsing traces left behind by the web surfer are relevant to the preference of the surfer. There were several studies tried to find a method to elicit surfers’ preference based on the browsing traces left behind by the web surfer. Various attributes of a product or service are presented in a web page, solicited surfers’ viewing traces were saved in the log files. Some researchers examined three pieces of information obtainable from the log files of the web server — (1) the order in which web pages were browsed, (2) the duration a surfer spent on each web page, and (3) the number of visited nodes which present information relevant to a particular attribute. The order of viewing is tied to how a web page is organized. The viewing duration is affected by many factors, such as connection speed, and/or side activities of the viewer. Therefore, this research tries to find if the number of visited nodes alone is highly correlated with surfers’ preference. Since the confounding factors attributed by the web organization and surfers’ anomalies have been removed, this research may provide some theoretical foundation of how to elicit surfers’ preference in reality. In order to accurately record every node a surfer has visited, a mechanism was developed to overcome the misplaced log caused by using back/forward button of the browser. Log anomalies caused by browser cache and proxy server are also resolved by this mechanism. A field study web site was set up and over one thousand surfers were elicited to view the content. This web site presented seven product attributes of the cellular phone service. Basic information about the viewers was acquired for gifts drawing after sufficient information was gathered. It has been found that the traces, represented by the number of nodes visited, are correlated with the surfers’ preference.