- 學位論文
利用網路服務發展智慧型個人化入口網站
Using Web Services to Develop Intelligent Personalized Portal
摘要
隨著系統整合範疇的擴大，整合也更為困難。傳統的分散式運算技術，因為制定規格上的限制，已無法滿足企業的需求。當今企業需要的是一種可以跨平台、跨程式語言、開發成本低、開發時間快的分散式運算技術。而Web Services就是基於這個目標所制定的業界標準。此外，傳統的資訊系統雖然具有提供資訊的能力，但其所提供的資訊無法滿足特定使用者的需求。為了克服這個難題，個人化將是一種可能的解決方法。 本研究利用Web Services與貝氏網路建立一個智慧型個人化入口網站系統， 該網站可以透過Web Services與其他異質系統相互連結，將各系統的資料加以彙整，並利用貝氏網路所實做的個人化引擎，把所得到的資訊依照使用者的偏好加以整理與排序。系統會將使用者的瀏覽歷程回傳給個人化引擎以調整貝氏網路的機率值，確保系統永遠使用最新的個人偏好值以滿足使用者的需求。 本研究依據前述的概念，本研究開發一個實證系統，用以比較各網路書店的書籍售價。這些網路書店可能存在於任何地方，使用不同的Web Services或是資訊系統，並且使用不同的貨幣。本入口網站必須使用一種方法用以存取這些網路書店，並且取得即時匯率使得所有的產品售價可以用同一種幣值顯示。利用貝氏網路將所得到的書籍資訊依照使用者的偏好加以整理，能讓使用者用較短的時間找到自己偏好的書籍。而此個人化處理會隨著使用者的偏好持續的調整。
並列摘要
As the scope of system integration is getting bigger and bigger, it is more difficult to integrate. The traditional distributed computing architectures are limited by their specifications and cannot satisfy the need of enterprises any more. Today, enterprises need a new distributed computing architecture with the ability of cross-platform, independence from programming language, low developing cost, and short developing time. Web Services is the new standard with these characteristics. Besides, traditional information systems are capable of providing information for all users, but not for a specific user. In order to overcome this issue, personalization is the key. This research uses Web Services and Bayesian Network to develop an intelligent personalized portal system. The portal could connect with other homogenous systems using the approach of Web Services, collect information from these systems, and sort information by a user preference with the personalization engine developed by the concept of Bayesian Network. The system sends back the browsing history of the user to the personalization engine to adjust the probabilities of Bayesian Network. This action could make sure that the system uses the newest preference of the user to always satisfy his/her need. According the concept proposed, this study develops a prototype portal which could compare the price of books among bookstores on internet. These bookstores might locate anywhere, have different Web Services or information systems, and use different currencies, so the portal has to find a way to access them and get the real-time currency exchange rate in order to show all the prices in the same currency. Using the Bayesian Network to sort the order of books with the user preference could make the user search more preferred books with much less time. This personalized process will continue to adjust according to the user’s preference.