A Study of Weighted Association Rules for Rapid Web-page Linkage
| Autor: | Shu-Huai Wang, 王肅惠 |
|---|---|
| Rok vydání: | 2004 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 92 Data mining technology can extract constructive and relevant rules from large amounts of data. If we employ the Standard Association Rule calculation method in analyzing Internet-surfing behavior, the resulting rules are likely to show deviations caused by website format. I.e., many websites are structured in a layered manner. If we only calculate the number of hits received by each web page, the pages closer to the homepage will naturally register a higher level of popularity; yet these pages are not necessarily the most-needed by the user. This article proposes an improvement on the above method of evaluating a web page’s significance. The proposed approach utilizes a new Least Weighted Association Rule calculation method, a new Weighted Value calculation method, an enhanced calculation method with regard to registered hits, and additionally considers the amount of time spent on a web page. Then based on this weighted value and the popularity of a given page, the Association Rule calculation method is applied to web logs to identify the web pages considered by users to be genuinely useful and interesting. This allows the website designer to better understand which pages attract users and use index page style web pages to incorporate more links, thus improving on the traditional, hard-to-navigate websites. Users are able to locate the page they need more efficiently. Experiments show that using the Weighted Association Rule to perform mining analysis will result in more trans-level rules due to the absence of the old linking rules derived from Standard Association Rules. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|