資訊管理學報
周世傑;吳政穎;
頁: 83-96
日期: 2009/01
摘要: 現今網路的搜尋引擎主要都是設計給大眾做資訊檢索，對於不同背景與需求的使用者，當提供一個相同的查詢句，所得到的搜尋結果會都是相同的大量網頁，這使得個人化搜尋的需求越來越高。使用者興趣檔描述了一個特定使用者的興趣，在資訊檢索的系統裡，通常用來幫助搜尋引擎提供個人化的搜尋結果，或者應用在推薦系統上。當今大部分的使用者興趣檔，是反映使用者長期的資訊需求，而非單次檢索的資訊需求。 本研究提出了一套方法，應用了概念擷取的技術，來幫助使用者建立一種短期的使用者興趣檔，希望藉由相關回饋來擷取出一個向量空間模型，用來代表使用者單次檢索的資訊需求，以幫助系統檢索出，與使用者需求相關的網頁給使用者瀏覽，以降低使用者的瀏覽時間，提高檢索的效率。
關鍵字: 使用者興趣檔;資訊檢索;資訊需求;概念擷取;向量空間模型;
Abstract: Nowadays, the search engine has been designed to retrieve information. With the same query, the users will get the same result which usually is a mass amount of web pages although the users' interests might be different. Therefore, enhancement on personalized search is always needed. The user profile which depicts the user's search interest has been applied to assist the user in the searching of personalized information in searching engine. Contemporary user profile mostly has been designed to reflect the user's long-term information interest. This study has proposed a method to develop the user profile which could reflect the user's short-term information interest. In the method, the information of relevance feedback has been utilized to extract the user's concepts of interest. These concepts of interest are then represented in a vector space model as the user profile for machine calculation. With the application of the vector space model of interest concepts, the web pages of the user's interest could be better retrieved to enhance retrieval efficiency.
Keywords: User Profile;Information Retrieval;Information Need;Concept Extraction;Vector Space Model;
瀏覽次數: 11018 下載次數: 3182
引用 導入Endnote
頁: 83-96
日期: 2009/01
摘要: 現今網路的搜尋引擎主要都是設計給大眾做資訊檢索，對於不同背景與需求的使用者，當提供一個相同的查詢句，所得到的搜尋結果會都是相同的大量網頁，這使得個人化搜尋的需求越來越高。使用者興趣檔描述了一個特定使用者的興趣，在資訊檢索的系統裡，通常用來幫助搜尋引擎提供個人化的搜尋結果，或者應用在推薦系統上。當今大部分的使用者興趣檔，是反映使用者長期的資訊需求，而非單次檢索的資訊需求。 本研究提出了一套方法，應用了概念擷取的技術，來幫助使用者建立一種短期的使用者興趣檔，希望藉由相關回饋來擷取出一個向量空間模型，用來代表使用者單次檢索的資訊需求，以幫助系統檢索出，與使用者需求相關的網頁給使用者瀏覽，以降低使用者的瀏覽時間，提高檢索的效率。
關鍵字: 使用者興趣檔;資訊檢索;資訊需求;概念擷取;向量空間模型;
Applying Relevance Feedback to Develop the Conceptual User Profile to Assist the User in Web Search
Abstract: Nowadays, the search engine has been designed to retrieve information. With the same query, the users will get the same result which usually is a mass amount of web pages although the users' interests might be different. Therefore, enhancement on personalized search is always needed. The user profile which depicts the user's search interest has been applied to assist the user in the searching of personalized information in searching engine. Contemporary user profile mostly has been designed to reflect the user's long-term information interest. This study has proposed a method to develop the user profile which could reflect the user's short-term information interest. In the method, the information of relevance feedback has been utilized to extract the user's concepts of interest. These concepts of interest are then represented in a vector space model as the user profile for machine calculation. With the application of the vector space model of interest concepts, the web pages of the user's interest could be better retrieved to enhance retrieval efficiency.
Keywords: User Profile;Information Retrieval;Information Need;Concept Extraction;Vector Space Model;
瀏覽次數: 11018 下載次數: 3182
引用 導入Endnote