請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/9338
標題: | 比較並結合基於內部和網路文件之查詢擴展方法 Comparing and Combining Query Expansion Approaches based on Local and Web Documents |
作者: | "Chien-Hung, Chen" 陳建宏 |
指導教授: | 李秀惠 |
關鍵字: | 虛擬相關回饋,關鍵字擴充,維基百科, Blind Relevance Feedback,Query Expansion,Wikipedia, |
出版年 : | 2009 |
學位: | 碩士 |
摘要: | 關鍵字擴展是增進資訊檢索效能的重要技術。依據擴展的文件集,可分為兩類,一類是基於內部文件集的關鍵字擴展,另一類則是基於網路文件集。
之前的研究顯示,基於內部文件集的擴展方法在增進檢索效能不佳的查詢上有瓶頸,然而,有人提出解決此問題之方法,是選用外部的文件集做為查詢擴展文件集,像是網路文件。 關於這兩類擴展方法的比較,我們的實驗結果顯示,的確在增進檢索效能不佳的查詢上,基於網路文件集-維基百科的關鍵字擴展有較好的效能,至於基於內部文件集而實行的關鍵字擴展,則在其他不同的查詢問題上有較好的表現。因此,我們提出將兩類不同文件集產生的擴展關鍵字列表作結合的方法,並評量其檢索效能的表現。大致上,我們擴展關鍵字的結合方法產生了較好的結果;但嚴謹的來說,我們只說我們的方式提供了平衡的結果。 Query expansion is an important technique to improve search capability in information retrieval. According to expansion collection, there are two types of query expansion. One is query expansion performed on local documents and another is performed on web documents. The previous research found the method which is based on local documents has bottleneck on poorly performing topics, called hard topics. However, others propose to improve poorly performing topics is exploiting text collections other than the target collection such as the web. Regarding our comparison of these two types of query expansion, our result shows query expansion based on web resource, which is Wikipedia, indeed has better performance on hard topics. As for query expansion performed on local documents, it has better performance on other topics. Therefore, we propose a combined method to integrate two ranked lists of terms expanded by these two types of query expansion, and evaluate the corresponding search performance. Roughly speaking, our combined query expansion methods produce better performance. However, to view it in a strict way, our methods provide balanced results. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/9338 |
全文授權: | 同意授權(全球公開) |
顯示於系所單位: | 資訊工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-98-1.pdf | 513.55 kB | Adobe PDF | 檢視/開啟 |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。