本研究利用HistCite™ 軟體 (Garfield, Pudovkin, & Istomin, 2003a, 2003b; see also Garfield, 2004)製作歷史圖表(historiography)，使Conrad Hal Waddington的作品(oeuvre)、引用這些作品的文獻以及引用連結(citation linkage)形成的網絡依據出版的時間順序呈現在圖形上，同時以NetDraw程式的 Newman–Girvan演算法(Girvan & Newman, 2002)確認在網路上具有高中介性的節點子群體。利用這個方式可以呈現研究狀態的脈絡並且發現連貫的研究串流(research stream)，以便歷時與跨領域地追蹤研究的想法和方法的擴散情形。
information visualization
A modified approach to algorithmic historiography is used to investigate the changing influence of the work of Conrad Hal Waddington over the period 1945–2004.
Much of this “citations as-indicators” work falls under the heading of evaluative bibliometrics and science & technology indicators (see, e.g., Moed, 2005).
Citation data can also be used to trace the diffusion of ideas and methodologies over time and across knowledge domains. They offer a window through which to view the development, growth, and decline of research fields and the impact or influence of both individual works and individual researchers, represented as bodies of cited work (oeuvres; see, e.g., Börner, Chen, & Boyak, 2003; De Mey, 1992; White & McCain 1989, 1997).
There are a number of different approaches to creating and visualizing longitudinal citation networks based on large document collections—examples include mapping of cocitation cluster strings (Small, 1977; Small, 2006; Small & Greenlee,1986; Small & Greenlee, 1989), textual and cocitation data mining (Chen, 2006), variable timelines based on core documents (Morris, Yen, Sheng, & Asnake, 2004), and main path analysis in temporal citation networks (Carley, Hummon, & Harty, 1993; Hummon & Doreian 1989; Hummon, Doreian, & Freeman, 1990). Related to the last approach is the general idea of algorithmic historiography, supported by the HistCite™ software (Garfield, Pudovkin, & Istomin, 2003a, 2003b; see also Garfield, 2004).
The associated historiographs are acyclic graphs—visualizations of the citation linkages among a subset of the documents that are all cited above a researcher-specified citation threshold. As in main path analysis, the networked documents are arranged in temporal publication sequence (older to younger) to show how research streams develop as newer work incorporates the information in older cited work and is, in turn, cited itself.
Published analyses using HistCite™ include a range of topics in information science (Garfield, 2004), networks associated with the discovery of DNA (Garfield, Pudovkin, & Istomin, 2003b), neural networks (McCain, 2005a), medical informatics (McCain & Silverstein 2006), and global environmental change (Janssen, Schoon, Ke, & Börner, 2006). In addition, Garfield has posted almost 400 analyzed document sets on his Web site (Garfield, 2006).
The challenge in tracing change over time bibliometrically is to retain the ability to see the world as it was at some particular time in the past and not just cumulatively, from the perspective of today.
Timeline visualization, main path analysis, and historiographic mapping have typically been applied to aggregate topical or journal-bound data sets with the temporal aspect being provided by the arrangement of the highly cited works.
The cumulative aspect of historiographic mapping becomes a particular problem if one wants to focus on a particular author rather than a broad topic or journal run as the basis for building the document set. The data set to be studied would, naturally, focus on the author’s oeuvre and the literature that cites it. Thus, the publications cited above any useful citation threshold will primarily be those of the author, with only the most prominent non-oeuvre-based citing articles visible in the network.
In this article, I demonstrate one way that historiographic mapping using Garfield’s HistCite™ software can be adapted to capture the impact of an author’s work. It preserves the context of the state of research at various points in time and allows identification of coherent research streams within a larger citing/cited network. This is accomplished by modifying and enhancing the data input, focusing on separate, successive periods, and using social network analysis software to build and analyze the historiograph.
Within-network research themes were identified usin NetDraw’s Newman–Girvan algorithm. This algorithm (Girvan & Newman, 2002) identifies subgroups of nodes within a network that have “high-betweenness centrality” (see Chen, 2006).
Waddington’s work is differentially cited by two distinct research communities. In Figure 1, it can be seen that on the left is a single, densely connected subnetwork focusing on canalization/genetic assimilation. To the center and the right, there are four subnetworks relating to experimental embryology and embryonic induction: amphibian embryology, theoretical models and experimental embryology, embryology of Drosophila, and embryonic induction.
A modified approach to algorithmic historiography is used to investigate the changing influence of the work of Conrad Hal Waddington over the period 1945–2004.
Much of this “citations as-indicators” work falls under the heading of evaluative bibliometrics and science & technology indicators (see, e.g., Moed, 2005).
Citation data can also be used to trace the diffusion of ideas and methodologies over time and across knowledge domains. They offer a window through which to view the development, growth, and decline of research fields and the impact or influence of both individual works and individual researchers, represented as bodies of cited work (oeuvres; see, e.g., Börner, Chen, & Boyak, 2003; De Mey, 1992; White & McCain 1989, 1997).
There are a number of different approaches to creating and visualizing longitudinal citation networks based on large document collections—examples include mapping of cocitation cluster strings (Small, 1977; Small, 2006; Small & Greenlee,1986; Small & Greenlee, 1989), textual and cocitation data mining (Chen, 2006), variable timelines based on core documents (Morris, Yen, Sheng, & Asnake, 2004), and main path analysis in temporal citation networks (Carley, Hummon, & Harty, 1993; Hummon & Doreian 1989; Hummon, Doreian, & Freeman, 1990). Related to the last approach is the general idea of algorithmic historiography, supported by the HistCite™ software (Garfield, Pudovkin, & Istomin, 2003a, 2003b; see also Garfield, 2004).
The associated historiographs are acyclic graphs—visualizations of the citation linkages among a subset of the documents that are all cited above a researcher-specified citation threshold. As in main path analysis, the networked documents are arranged in temporal publication sequence (older to younger) to show how research streams develop as newer work incorporates the information in older cited work and is, in turn, cited itself.
Published analyses using HistCite™ include a range of topics in information science (Garfield, 2004), networks associated with the discovery of DNA (Garfield, Pudovkin, & Istomin, 2003b), neural networks (McCain, 2005a), medical informatics (McCain & Silverstein 2006), and global environmental change (Janssen, Schoon, Ke, & Börner, 2006). In addition, Garfield has posted almost 400 analyzed document sets on his Web site (Garfield, 2006).
The challenge in tracing change over time bibliometrically is to retain the ability to see the world as it was at some particular time in the past and not just cumulatively, from the perspective of today.
Timeline visualization, main path analysis, and historiographic mapping have typically been applied to aggregate topical or journal-bound data sets with the temporal aspect being provided by the arrangement of the highly cited works.
The cumulative aspect of historiographic mapping becomes a particular problem if one wants to focus on a particular author rather than a broad topic or journal run as the basis for building the document set. The data set to be studied would, naturally, focus on the author’s oeuvre and the literature that cites it. Thus, the publications cited above any useful citation threshold will primarily be those of the author, with only the most prominent non-oeuvre-based citing articles visible in the network.
In this article, I demonstrate one way that historiographic mapping using Garfield’s HistCite™ software can be adapted to capture the impact of an author’s work. It preserves the context of the state of research at various points in time and allows identification of coherent research streams within a larger citing/cited network. This is accomplished by modifying and enhancing the data input, focusing on separate, successive periods, and using social network analysis software to build and analyze the historiograph.
Within-network research themes were identified usin NetDraw’s Newman–Girvan algorithm. This algorithm (Girvan & Newman, 2002) identifies subgroups of nodes within a network that have “high-betweenness centrality” (see Chen, 2006).
Waddington’s work is differentially cited by two distinct research communities. In Figure 1, it can be seen that on the left is a single, densely connected subnetwork focusing on canalization/genetic assimilation. To the center and the right, there are four subnetworks relating to experimental embryology and embryonic induction: amphibian embryology, theoretical models and experimental embryology, embryology of Drosophila, and embryonic induction.
沒有留言:
張貼留言