透過您的圖書館登入
IP:18.118.1.158
  • 學位論文

以大眾分類法為基礎之網站內容分類架構—以社群書籤網站為例

A Classification Framework of Website Content Based on Folksonomy in Social Bookmarking

指導教授 : 皮世明

摘要


自動化文件分類技術在知識管理領域應用相當廣泛,傳統上自動化文件分類技術主要以兩大方向進行。第一種方向是以關鍵字為基礎的分類方法,例如早期的TFIDF以及近幾年應用廣泛的支援向量機方法(SVM)等。但是以關鍵字為基礎的分類分法卻擁有關鍵字語意不清問題。第二種方向則是以語意分析為基礎的分類方法,早期是以關鍵字的語意分析為主,而近幾年有專家學者提出以本體論解決語意問題,但本體論在建構上卻有專家知識領域代表性的疑慮,較無客觀的建構方法。因此,大眾分類法(Folksonomy)便是在Web2.0的衝擊與眾多分類問題之情況下衍生的分類概念,以廣大的使用者取代以往專家定義資訊的現象。但大眾分類法仍是以關鍵字為基礎,依舊有語意上的問題。故本研究提出一個大眾分類權重機制,應用於社群書籤網站。期望能解決大眾分類在語意同義字及分類效果不佳的問題。 為了改善大眾分類效果之問題,本研究擬提出一個大眾分類的權重機制應用於社群書籤網站上。首先在使用者訂定個人書籤時,會自動收集個人書籤之標籤關鍵字(Tag)。接著將標籤關鍵字進行斷詞處理,再利用WordNet詞彙庫查詢同義詞之相關詞彙。最後運用TFIDF詞彙計算的概念,計算出同義詞的分類權重值及進行細項的調整後即完成分類的動作,並將分類結果列表供使用者查詢。 研究結果顯示,本研究提出的大眾分類權重機制,有效縮減標籤分類的數量達百分之三十以上,且明顯改善標籤分類之品質及增加使用者的滿意度。表示本研究提出的大眾分類權重機制,可有效的改善大眾分類中語意同義字以及分類效果不佳等問題。

並列摘要


Document classification technique applies extensively in knowledge management and enterprise. Automation document classification has focused on two dimensions. The first domain is keyword-based that is based on TFIDF in the early time, and develop into SVM for the modern way. However, the keyword-based automation document classification has problems of semantic. The second classification is semantic-based that has focused on keyword-based problem-solving. Many researchers propose that ontology-based classification can solve keyword-based semantic problem. But need to concern with how to build ontology and the representation of defining the domain of expert knowledge. Due to these dubious interpretation, Folk Classification (Folksonomy) is produced of Web2.0 conception, and based on keyword-based. This paper proposed a Folk Classification Bookmark System that com combined WordNet and TFIDF classification method. We expect the Folk Classification Bookmark System can solve the semantic and classification problem. This research proposes the mechanism of Folk Classification Bookmark System that integrates WordNet and TFIDF technologies. Users can define the tag by themselves. After separating the keyword from tags, the system will find the synonym from WordNet. Finally the synonym would use TFIDF to classify and user can query or browse from the “keyword”. On the research, results of this study show that Folk Classification achieves 30% or higher data reduced rate. The result of the classification promotes the classified data quality and increase user satisfaction. On the conclusion, this study proposes the mechanism of Folk Classification, and shows that Folk Classification is capable to improve of synonym and classified problem.

參考文獻


26. 游子賢,「通俗分類標記之自動組織」,國立台灣大學資訊管理學系研究所未出版碩士論文,2006。
3. Golder, S., and Huberman, B. A., “Usage patterns of collaborative tagging systems,” Journal of Information Science,” 32(2), 2006, pp. 198-208
5. Gregg, D.G. and Walczak, S., “Auction advisor: An agent-based online-auction decision support system,” Decision Support System, 41(2), 2006, pp. 449-471.
7. Kobayashi, M., and Takeda, K., “Information Retrieval on the Web,” ACM Computing Surveys, 32(2), 2000.
11. Miller, G. A., “WordNet: A Lexical Database for English,” Communication of the ACM, 38(11), 2005.

被引用紀錄


張慈育(2009)。大眾標記法應用於考古文物描述之研究:以國小五年級學生標記十三行博物館文物為例〔碩士論文,國立臺灣大學〕。華藝線上圖書館。https://doi.org/10.6342/NTU.2009.03020
吳佳典(2010)。以協同合作模式建構研究者知識之研究〔碩士論文,國立臺灣師範大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0021-1610201315205072

延伸閱讀