Title: 一個針對行動通訊使用者所設計的搜尋引擎
A Search Engine Designed for Mobile User
Authors: 洪健洲
Chien-Chou Hung
羅濟群
hi-Chun Lo
資訊管理研究所
Keywords: 搜尋引擎;行動通訊;行動計算;情境資訊處理;search engine;mobile communication;mobile computing;context processing
Issue Date: 2002
Abstract: 搜尋引擎是網際網路世界裡的殺手級應用之一,若是能將搜尋引擎建置到行動通訊的領域,將可方便使用者取得無數的網際網路資源。不過,現行的搜尋引擎大都是設計給需要長時間、專注且位置不動的使用者使用,對於經常移動,無法專注於搜尋動作之上的個人行動通訊使用者而言,這樣的搜尋引擎是無法幫助他們方便地獲得所需要的資訊。日本NTT DoCoMo的MIS系統,則是引進了適地性服務(Location-based Service)的概念,提供使用者各項與位置有關的搜尋服務。然而,使用者的情境資訊,代表著使用者使用搜尋服務的當時的環境,很顯然地不單只有位置一種,位置、時間甚至使用者偏好,都可以用以幫助搜尋引擎產生更符合使用者需求的搜尋結果。另外,粗糙地將各種位置資訊轉換成經緯度表示,因尺度不同所導致的誤差問題亦無法解釋。因此,本研究綜合位置、時間以及使用者偏好等三種情境因子,定義使用者與文件的情境,提出一套以分類樹(Ontology Tree)的計算為基礎的通用機制,以計算文件與使用者情境的相似度,用以作為修改搜尋結果的排序(Ranking)依據。實驗結果證明,這樣的機制,能夠有效地將符合使用者情境的文件,排在搜尋結果的較前面,使得搜尋引擎更加地適合給行動通訊使用者使用。
Search engine is one of the killer applications in Internet. It will be convenience for user to access countless World Wide Web resource to build the search engine in the mobile communication environment. But, mostly, traditional search engine are designed for focused attention and stationary use. For those mobile communication users who move regularly and unable to concentrate on searching, this kind of search engine cannot easily help them get what they need. The NTT DoCoMo MIS system in Japan adopted the concept of location-based service, provide various searching services related to location. However, the user context information represents the environment and possible intention of user, and obviously there is not only one: location. The location, time, and even user preference could be used to help the search engine to produce more suitable search result to user need. The deviation produced by converting longitude and latitude from different value scale can not be explained, too. Therefore, our research combine location, time, and user preference these contexts factor to define user and document context, and purposed a generalized mechanism based on ontology tree calculation, to compute the similarity between document and user context, in order to adjust the original search result ranking. Experiments results show that this mechanism can effectively pick up those documents that are related to user context, and arrange to the top of search result. Make the search engine more suitable for mobile communication users.
URI: http://140.113.39.130/cdrfb3/record/nctu/#NT910396019
http://hdl.handle.net/11536/70292
Appears in Collections:Thesis