標籤使用行為分析應用在專業知識之自我學習

The tagging behavior analysis for hard contents and its applications in self-learning

蔡麗珍(Li-Chen Tsai)

指導教授 ： 黃雪玲 ； 唐國豪

國立清華大學/工學院/工業工程與工程管理學系/博士(2014年)

摘要

自從嵌入式社會計算及Web2.0環境的普及，越來越多的社會計算應用，例如標籤，被採用於現今網路環境。標籤可以從資訊內容中擷取概念並做為瀏覽導航提示(navigational cues)，並使得使用者透過標籤找到有意義且相關的資訊。這對於專業領域生手理解以不同專業程度撰寫之正規的學術或科學領域的文章尤其重要。
本研究探討三項有關標籤技術的議題。首先，研究探討專家與生手族群挑選標籤的差異，並利用相似性與相關性量測方法探討標籤品質問題。實驗結果顯示，相對於生手來說，專家對於學術與科學相關的文章可以產生較具代表性與一致性的標籤；同時，也證明專家挑選出的標籤確實能反應出對資訊內容較佳的理解。
研究的第二部分則討論專家與生手兩族群，受社會影響的狀況下標籤收斂的分佈變化；並且比較三種方法衡量在兩族群中標籤收斂的分佈變化。實驗結果表示，專家群組的標籤收斂速度較生手為佳。此外本研究亦提出 “one-bit comparison”方法，該方法能夠有效的自所有標籤中，辨識出已具備高度共識、成熟的專家標籤。
研究最後一部分則是使用成熟且高品質的標籤，驗證標籤應用於自我學習的有效性。實驗探討是否可以透過專家萃取的標籤輔助學習以增加學生的學習表現。實驗結果表示專家選擇的標籤能夠幫助學生對內容有更好的理解，對生手具備提供自我學習的輔助作用。
本篇論文探討在排序標籤分配的共識收斂過程中，專業知識所扮演的角色。研究結果支持以標籤為基礎的學習模式，並且提供在WEB 2.0環境中一包含標籤的介面設計概念與工具。

關鍵字

標籤行為 ； 標籤品質 ； 排序標籤分配 ； 收斂 ； 自我導向學習 ； 專業知識水準

並列摘要

Since the advent of Web 2.0 and embedded social computing, there has been a widespread adoption of social computing applications such as tags. Tags can extract concepts from content and act as navigational cues that enable users to find meaningful and relevant information. This is especially important for domain novices in understanding formal academic or scientific articles written at varying domain expertise levels.
In this study, three topics focusing on tag technology are discussed. First study elicited differences in tag assignments by Expert and Novice groups, and discussed tag quality problems in terms of similarity and relevance measures. The results show that experts can provide a more consistent and representative set of tags for academic and scientific documents than novices can generate, suggesting that tags chosen by experts reflect better understanding of the content.
The second part of the study discusses the convergence variation of tag distributions that are affected by the social influence of a group of domain experts or a group of domain novices. This study compares three measures of the convergence rate of tagging behavior in Expert and Novice groups. The results show that the convergence rate of tagging behavior was better in the Expert group than it was in the Novice group. The one-bit comparison proposed by this research can accurately distinguish mature tags generated by experts with high consensus from other tags.
The final part validated the effectiveness of using mature and high quality tags to facilitate self-directed learning. The experiment measured whether or not students can increase learning performance through these tags that had been extracted by domain experts. The result revealed that tags chosen by experts helped students’ better understanding of the content.
This dissertation investigates the roles of expertise during convergence of consensus of a rank-ordered tagging distribution. The results support tag-based learning and provide insights and tools toward the design of interface involving tags in the Web 2.0 environment.

並列關鍵字

tagging behavior ； tag quality ； rank-ordered tagging distribution ； convergence ； self-directed learning ； expertise level

參考文獻

Abel, F., Henze, N., & Krause, D. (2008). Exploiting additional Context for Graphbased Tag Recommendations in Folksonomy Systems. In: Int. Conf. on Web Intelligence and Intelligent Agent Technology, Sydney, Australia. ACM Press.

Alexander, B. (2006). Web 2.0: a new wave of innovation for teaching and learning. Educause Review, 41(2), 32-44.

Allen, B.L. (1991). Topic knowledge and online catalog search formulation. Lib. Quart., 61(2), 188-213.

Anderson, J.R. (1984). On the merits of ACT and information-processing: A response to Wexler's review. Cognition, 8, 73-88.

Angus, E., Thelwall, M. & Stuart, D. (2008). General patterns of tag usage among university groups in Flickr. Online Information Review, 32(1), 89-101.