文獻內容探勘工具- CATAR-之發展和應用
No Thumbnail Available
Date
2011-04-??
Authors
曾元顯
Journal Title
Journal ISSN
Volume Title
Publisher
國立台灣師範大學圖書資訊研究所
Graduate Institute of Library and Information Studies
Graduate Institute of Library and Information Studies
Abstract
近年來，隨著各種文獻資料庫的蓬勃發展，以及研究績效評鑑的需求，文獻內容探勘與分析的應用，特別是根據科學計量分析的方法，越來越受重視。然而相關的軟體工具，卻不敷使用。本研究發展一套名為CATAR 的內容探勘工真，提供研究者進行科學計量等分析之用。其可進行概要分析，針對待分析文件的各項欄位(如作者、機構、國家、類別、出處、年代、被引用情況等等)進行攝取、統計、運算與排序，也可根據書目對或共現字進行分解分析，擷取待分析文件中的主題，並與各項欄位進行多樣的交叉分析。本文除簡介此工具使用的技術與方法，更詳細說明其使用方式，並以具體案例，展示其分析功能。這些案例，包括概要分析的應用、書目對與共現字的主題歸類效果、及主題趨勢演變分析等。這些分析，可提供由下而上、資料驅動、證據為主的研究線索，供後續的驗證與探討。此工具己放在網路上提供免費使用，以擴散科學計量方法評量績效的應用，或是在課堂上供教師進行領域的回顧與前膽、在課後供學生自行分析研究主題的趨勢與方向，以協助擬定後績的研究題目。
With the advent of various publication databases and the demands for performance evaluation of research output, content analysis, especially the scientometricsanalysis, receives more and more attention. However, there exist very few tools that can aid non-technicians to do the analysis. This article reports the development of a simple-to-use content mining tool named CATAR to help various scientometrics analyses. It provides overview analysis that can extract various fields from the documents to be analyzed and sort the documents by fields in meaningful ways. It also provides breakdown analysis which identifies the topicsrevealed in the documents and provides cross-tabulation analyses between the topic and field. In addition to introducing the techniques and methods used in this tool , the detailed usage and some case studies are described. The described cases include an overview analysis application , topic clustering based on ibliographic coupling and co-word analysis, and a topical trend analysis. These analyses offer evidence-based, data-driven, bottom-up research clues for further verification and exploration. This tool has been opened on the Web for free use in the hope to promote the application of scientometrics to performance valuation, to conduct the overview and foresight of a field by teachers in classes, or to allow self-exploration of research trends by students after classes for etermination of their research topics.
With the advent of various publication databases and the demands for performance evaluation of research output, content analysis, especially the scientometricsanalysis, receives more and more attention. However, there exist very few tools that can aid non-technicians to do the analysis. This article reports the development of a simple-to-use content mining tool named CATAR to help various scientometrics analyses. It provides overview analysis that can extract various fields from the documents to be analyzed and sort the documents by fields in meaningful ways. It also provides breakdown analysis which identifies the topicsrevealed in the documents and provides cross-tabulation analyses between the topic and field. In addition to introducing the techniques and methods used in this tool , the detailed usage and some case studies are described. The described cases include an overview analysis application , topic clustering based on ibliographic coupling and co-word analysis, and a topical trend analysis. These analyses offer evidence-based, data-driven, bottom-up research clues for further verification and exploration. This tool has been opened on the Web for free use in the hope to promote the application of scientometrics to performance valuation, to conduct the overview and foresight of a field by teachers in classes, or to allow self-exploration of research trends by students after classes for etermination of their research topics.