Preferred Traversal Path Mining with HITS
| Autor: | Ying-Lin Lin, 林盈霖 |
|---|---|
| Rok vydání: | 2009 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 98 Web usage mining can be used to discover useful information hidden in web log data. However, many previous algorithms do not consider the structure of web pages, but regard all web pages with the same importance. On the other hand, web structure mining often utilizes HITS (hypertext induced topic selection) to rank web pages. HITS algorithm ranks web pages by analyzing the in-degree and out-degree of a webpage. The concept of preference is that if there are many different options to leave a page, the options that are selected most frequently and the next page viewed reveal user interest and preference. PNT (preferred navigation tree) is an algorithm that finds users’ preferred navigation paths. The main objective of this research is to extract users’ preferred traversal paths by using HITS values and PNT algorithm preferences as measures. This research introduces the Preferred Navigation Tree with HITS (PNTH) algorithm, which is an extension of PNT algorithm. PNTH algorithm uses the concept of PNT and takes into account the relationships among web pages using HITS algorithm. The experimental results show that PNTH algorithm is more accurate than PNT algorithm. Furthermore, the running time of PNTH algorithm and PNT algorithm are very close. PNTH algorithm is suitable for E-commerce applications such as improving web site design and web server performance. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|