碩士 / 國立高雄大學 / 資訊管理學系碩士班 / 101 / With explosive growth of the Internet, huge amount and complicate information have been aggregated on the web. Adaptive website has been considered a technique that can present the information that users needed by analyzing users’ behavior. However, users may have different needs at different times and most of recommended methods are not functioned as dynamic or time-dependent needs. In this paper, we propose a web page navigation recommendation approach which is based on reinforcement learning technique, and applies it to informational website. Five parameters are considered and included in the recommendation approach, which include clicks of the page, time that users stay at the page, paths to achieve the page, hierarchical level of the page, and the rank of the page. With reinforcement learning, the website adjusts the weight of five parameters automatically. This paper aim to reduce the paths that user needed to achieve the object page. According to the empirical evaluation results, it shows that the path length to object pages can be reduced and the recommendation that included clicks perform better performance than traditional method which only the parameter of clicks is considered.
| Identifer | oai:union.ndltd.org:TW/101NUK05396007 |
| Date | January 2013 |
| Creators | Yin-ling Tang, 唐銀伶 |
| Contributors | I-Hsien Ting, 丁一賢 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 67 |
Page generated in 0.0095 seconds