Please use this identifier to cite or link to this item:
Title: 自動分群搜尋引擎之使用者評估研究
User-based Study of Automatic Clustering Search Engines
Authors: 卜小蝶
Pu, Hsiao-Tieh
Chen, Sih-Ying
Keywords: 自動分群
automatic clustering
relevance ranking
earch engines
user study
Issue Date: 2007
Abstract: 隨著網路資源迅速成長,利用搜尋引擎檢索網路資源,也成為使用者最有利的檢索工具。然而,現今以相關排序為主的搜尋引擎,仍無法有效地過濾龐雜的檢索結果,反而容易造成使用者的困擾。自動分群搜尋引擎則提供使用者另一種選擇,以自動分群的方式提供使用者分門別類的群集主題,藉以改善檢索效益。 雖然自動分群技術之相關研究已行之有年,但仍缺少使用者方面之相關研究,因此本研究即嘗試以使用者觀點評估群集架構及其使用性。本研究設計兩階段實驗,讓使用者實際參與任務,包含使用者自訂情境與研究者指定情境,並從中觀察使用者使用自動分群搜尋引擎之情形,再利用歷程分析紀錄使用者之檢索歷程,搭配問卷與訪談深入瞭解使用者主觀認知。研究中採用自建之實驗平台、檢索任務及評估問卷、訪談大綱、電腦螢幕錄製軟體等,以利研究之進行。並依據研究目的與問題獲取所需之研究素材,以觀察及訪談方式整理歸納受試者之檢索行為特性,及利用檢索歷程紀錄、訪談、評估問卷等方式,分析搜尋引擎之檢索效率與效益,最後則以訪談與評估問卷分析使用者滿意度。 研究結果發現,使用者使用自動分群搜尋引擎與相關排序搜尋引擎之最大差異在於,使用者過濾資訊時所花費的時間與心力,以及使用之時機皆有所影響;其次,自動分群搜尋引擎對使用者最大的幫助在於,有較佳的檢索效益、可以縮小檢索主題範圍、突顯重要概念、提供多維思考方向,並在簡單/封閉的問題有最佳檢索表現。最後,本研究提出改善自動分群搜尋引擎之建議,包括依據使用者所需提供常用之群集類別與組合、與自訂層級及其檢索結果數量等個人化的服務,或是參考人工分類以及使用者回饋等方式,以使用者為導向提升分群的品質。
This study tries to evaluate the structure and the usability of automatic clustering search engines based on the user’s perspectives. The rapidly increasing amount of Internet resources has made search engines one of the most important tools for searching and accessing Internet resources. The results of major searchengines found and presented are based on relevance ranking. They may not be able to effectively and efficiently filter the results since they have brought difficulties in terms of precisely locating what the user is seeking for because of the intimidating amount of the results found and presented. Automatic clustering search engines have offered the user a better option. They provide the user with the function of automatically clustering the categories by topics and may thus increase the effectiveness of searching. This study designs experiments that enabled the user to actually participate in the searching tasks of using automatic clustering search engines. The tasks that this study designs include self-defined tasks by the participants and the tasks that this study assigns. This study observes the participants’ behaviors during the process of using the automatic clustering search engines assigned. It also records the searching process by log to analyze the effectiveness of the search engines selected. Finally, it tries to understand the satisfaction of the user by interviews and evaluation questionnaire after the tasks assigned were completed.. The research tools that this study adopts include the experimental platform, searching tasks, evaluation questionnaire, interviews outline, and log analysis software. The results of this study suggests that the time spent and the efforts made by the participant on using the automatic clustering search engines selected are strongly different. The contexts in which the participant used the automatic clustering search engines are also highly different from using the search engines based on relevance-ranking. The results also showed that the automatic clustering search engines selected help enhance the effectiveness of the searching, narrow the topic scope of the searching, highlight key concepts, and provide diverse thinking for searching. Finally, this study provides suggestion to improve automatic clustering search engines. Various combinations of the clusters that the user frequently uses, personalized hierarchical clusters and self-defined numbers of the searching results presented, human-designed clusters and user’s feedback may be provided to enhance the quality of the clustering based on the user.
Other Identifiers: GN0693150072
Appears in Collections:學位論文

Files in This Item:
File Description SizeFormat 
n069315007201.pdf205.05 kBAdobe PDFView/Open
n069315007202.pdf1.81 MBAdobe PDFView/Open
n069315007203.pdf797.35 kBAdobe PDFView/Open
n069315007204.pdf637.79 kBAdobe PDFView/Open
n069315007205.pdf174.14 kBAdobe PDFView/Open
n069315007206.pdf532.81 kBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.