本研究設計並實作一個meta search代理人（Meta Search Agents，MSA )，以克服 一般搜尋引擎之不足處。本研究設計之MSA的五個功能模組與提出之演算法為：⑴ 查詢模組一使用者輸入欲查詢之關鍵字，並設定相關查詢條件；⑵資訊檢索模組一代 理人透過分派演算法（DispatcherAlgorithm)啟動符合資料來源類型的各個搜尋引擎 進行檢索，並擷取各檢索結果之網頁原始碼；⑶資訊萃取模組一透過特徵萃取演算法 (Features ExtractionAlgorithm)以萃取網頁原始碼中的重要標籤，再經超連結正規化 演算法（Hyperlinks Normal Form Algorithm)去除格式不合法之超連結；⑷資訊過濾模 組一使用個數與次數演算法（Occurrence Hit Algorithm)以計算各超連結之搜尋引擎指 向個數與超連結指向次數，並經過濾超連結演算法（Filter Hyperlinks Algorithm)移除 與關鍵字不相關之超連結，再使用關鍵字頻與位置（KeywordFrequency and Position) 演算法計算各超連結之分數；（5)資訊整合模組一代理人彙整各個超連結並以友善的書 籤式界面呈現，方便使用者點選與瀏覽代理人之檢索結果。本研究所設計並實作之 MSA具有高精確度、高回憶度與高效能的特性，並能降低使用者的資訊負荷。

In this paper, we design and implement a Meta Search Agents called MSA that overcomes the drawbacks of search engines. The MSA is able to consult many search engines for a single query at the same time by reducing the time spent on accessing multiple search engines. The MSA includes five main functional modules as follows. (1) Query Module — the interface of input query keywords and query conditions by users. (2) Information Retrieval Module — MSA sends the query keywords and query conditions to different search engines by the Dispatcher Algorithm. (3) Information Extraction Module — MSA extracts the important tags and delete the illegal hyperlinks by the Features Extraction Algorithm and the Hyperlinks Normal Form Algorithm. (4) Information Filtering Module — MSA ranks the query results by the Hyperlinks Algorithm and the Keyword Frequency and Position Algorithm. (5) Information Integration Module — the output options of collating results. The main contribution of the MSA is a metho d for reaching high recall and precision, and decreasing information overload to the users.