Using Chance Discovery and Web Mining for Personalized Advertisement Recommendation

• Main Authors: Chia-Chi Pan, 潘佳琪
• Other Authors: Sung-Shun Weng
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/77794010629964700284

碩士 === 輔仁大學 === 資訊管理學系 === 96 === Owing to the tremendous development of Internet leads to the popularity of Internet marketing service. However, not only the design of website itself will influence Internet marketing service, but also Web advertisement will be the important key to attract users’ attention. Hence, it can be seen that Web advertisement has become the important transmission and marketing medium for the web generations. Therefore, Web advertisement should no longer be broadcasted by a random way. It should be able to respond to users immediately and recommend the interesting items that the customers prefer to as well as satisfying items to them. Based on such concept, the recommendation systems thus generate. Traditional recommendation systems emphasize the accuracy in order to provide information of critical and particular items. But information of such systems will be consumed quickly and easily. Accordingly, our research applies the KeyGraph algorithm of chance discovery to construct the relationships between words and words and find out the nodes which have the qualities of chance discovery. We utilize these nodes to recommend novel items which connect with those keywords. But, sometimes such novel items seem to have incomplete relation which users can hardly identify its relation by intuition. Furthermore, we also apply the technique of Web mining to analyze the browsing behaviors of users, understand what kind of domain that users are interested in, and estimate the users’ interesting degree on the content of webpage as to extend the recommending scope of the Web advertisement. Our approach incorporates chance discovery with Web mining to keep recommendable list diversified and various, continue increasing the quantity of recommendable items, and provide the uninterrupted recommendable behavior.