層面分類結構應用於社會性標記之研究：以圖書作品為例

Abstract

社會性標記（social tagging）的發展已成為新一代網路應用的重要趨勢之一。然而，現有的許多標記系統所產生的標籤大多是缺乏結構化與組織的結果，因而可能難以提供具語意結構的、多面向的資訊，較無法有效支援資源的瀏覽或檢索。有鑑於此，本研究提出以層面分類作為改善標記品質的方法，嘗試將層面分類的概念運用在標記系統中，以探討層面分類結構和社會性標記相結合的可行性與應用成效。本研究選擇以圖書作品做為標記的資源對象，並討論以下問題：（1）透過層面分類結構進行標記與否，對於圖書標記結果是否會造成影響並產生差異？（2）當面對不同的圖書作品文類時，層面分類結構對於標記結果的影響是否因而有所差異？（3）層面分類結構的提供是否為使用者所需，又使用者對於該層面分類結構的看法或態度為何？ 本研究採取實驗法之設計，並透過自行設立之標記系統與問卷蒐集相關實證資料，以統計檢定分析測量結果。 研究結果發現：（一）在標籤集合特性方面，以層面分類結構進行標記有助於產生較多的標籤數量、較多重疊語意之詞彙、較高共識程度與聚集現象、以及較多面向之標籤類型；（二）在標記操作成本方面，層面分類結構的提供會導致標記者花費較多完成時間，且在標記過程中覺得較不順利；（三）在標籤自評結果方面，以層面結構進行標記者對於所建標籤的滿意度和正確度評價皆顯著偏低；（四）層面分類的提供對於標籤平均數量的影響不會因圖書作品文類而產生顯著差異；（五）比較小說與非小說文類之間的標籤結果，小說作品具有較少的相異標籤、較高的同義標籤相似度、以及較高之標籤使用集中程度；（六）多數標記者都認同層面結構之助益，但是他們對於未來繼續使用層面分類結構建立標籤的意願程度卻不高。 最後研究針對層面分類於標記之應用提出實務建議：（一）提供更具彈性化和多樣化的層面分類結構；（二）在標記系統介面上同步呈現依層面欄位輸入標籤後的資訊；（三）增加標記系統的易用性，以減少使用者的標記操作成本；（四）提供相關的額外系統輔助功能；（五）增加標記之動機與誘因，以鼓勵使用者多花費一些心力來產出較佳的標籤品質。

Recently, social tagging has been suggested as the supplement to traditional ways of information organization because it holds the promise of adding value to controlled vocabularies-based cataloging by enabling users to organize books or titles of interest with their own tags. However, several studies have questioned its effectiveness in information discovery and retrieval. The main assumption of the study is that tagging can be effectively mixed with a richer faceted classification scheme to support semantic browsing and searching. This study aims to compare faceted structure-based tagging with freely assigned tags of the same books and to understand the effect of different tagging modes on the resulting tag sets, particularly in the context of fiction and non-fiction works. An experiment was conducted with a factorial and randomized post-test control group design. The participating users were assigned to either experimental group or control group depending on the provision of faceted structure. All the participants of both groups were asked to tag 16 titles comprising 8 fiction and 8 non-fiction works. Research findings show that the the resulting tag sets in faceted structure-based tagging display more distinct tags, more number of assigned tags on average, higher tag similarity , higher convergence, and more facets. However, findings also reveal that tagging with faceted structure takes more time and effort, and the users feel less satisfied and less confident of the resulting tag sets. The last finding suggests that tagging can be effectively mixed with a faceted classification scheme to support book indexing and retrieval. Nevertheless, the faceted classification needs to be enhanced with more flexibility, greater diversity and easiness.