本研究利用社會網絡分析與科學地圖映射(science mapping)分析Scientometrics期刊2002到2008年發表的816筆論文。
針對Scientometrics期刊進行書目計量分析的相關研究，包括：Schoepflin and Glanzel (2001)將Scientometrics在1980、1989和1997年發表的論文分別進行歸類，發現科學政策(science policy)和科學社會學(the sociology of science)的比率在下降。Peritz and Bar-Ilan (2002)發現Research Policy和Social Studies of Science分別是1990和2000年Scientometrics論文引用的期刊次數最多的第三名和第四名。Chen, McCain, White, and Lin (2002)分析出1981到2001年間Scientometrics期刊的引用及共被引模式。Hou, Kretschmer, and Liu (2008)對2002到2004年間Scientometrics期刊上的作者合作網絡的結構特性進行分析。Dutt, Garg, and Bali (2003) 則分析Scientometrics期刊1978到2001年間論文資料上的國家、機構在主題上的分布。
本研究在816筆論文資料上共計發現57個國家，具有較大生產力的國家主要是歐洲國家。前十個較大生產力的國家裡，美國、比利時、西班牙、中國和德國都有相當快速的年增率，但印度的年增率是負的。生產力較大的國家的被引用次數也比較高。
為了國家間的研究合作情形，本研究提出相對合作強度(relative collaborative intensity, RCI)，這個測量方式整合了合作的國家數和合作的次數兩種指標，其公式如(3)所示：
假設(RCI)i是第i個國家的相對合作強度，其中CCi和CTi分別是這個國家合作的國家數和與其他國家合作的次數。在本研究裡，比利時是相對合作強度最高的國家，英國、荷蘭與美國則分居2到4名。
接下來將國家間的合作關係表現成網絡圖，圖形上最大的相連成分(connected component)共有37個國家。在這個相連成分上，比利時、英國和匈牙利之間都有很強的連結。
以機構來看，比利時的Katholieke Univ Leuven、匈牙利的Hungarian Academy Science和荷蘭的 Leiden Univ發表的論文數和被引用次數最多。
進一步分析前十個主要機構的被引用次數最多的前十筆論文資料，發現引用它們的論文主要來自圖書資訊學、電腦科學、資訊系統和跨領域應用(interdisciplinary applications)等領域。但台北醫學大學的一篇論文則被許多生物醫學領域的論文引用。
就論文的合作作者數來分析，本研究發現單一作者的論文有271篇，多位作者的論文有545篇，每篇論文平均有2.29位作者。Dutt, Garg, and Bali (2003) 研究1978-2001年間的論文，單一作者的論文占半數一上，平均合作作者數則為1.73。兩相比較之下，由多位作者的論文數和平均作者數增加的結果，能夠顯示Scientometrics期刊上的合作情形增多。
從引用的文獻分析Scientometrics的主題包括科學與技術的關係(the relationship between science and technology)、個人科學研究產出的量化指標(indexes to quantify an individual's scientific research output)、作者的合作現象(author collaborations)、共被引網絡(co-citation networks)、科學引響力以及國家富強(the scientific impact and wealth of nations)。
The purpose of this article is to use the methods of Social Network Analysis and Science Mapping to make an analysis on the 816 papers published in the international journal Scientometrics from 2002 to 2008.
The major tools used in this paper were TDA, NWB and Excel.
Börner (2006) discussed the mapping research on structure and evolution of science.
Börner, Penumarthy, Meiss, and Ke (2006) mapped the diffusion of information among 500 major U.S. research institutions based on the 20-year publication data set published in the Proceedings of the National Academy of Sciences (PNAS) in the years 1982-2001.
Boyack, Börner, and Klavans (2009) mapped the structure and evolution of chemistry research over a 30 year time frame based on Science (SCIE) and Social Science (SSCI).
Leydesdorff and Rafols (2009) made a global map of science based on the ISI subject categories.
For instance, Schoepflin and Glanzel (2001) found a decrease in the percentages of both the articles related to science policy and to the sociology of science by classifying the articles published in Scientometrics in the years 1980, 1989 and 1997.
Peritz and Bar-Ilan (2002) analyzed the papers published in Scientometrics in 1990 and 2000 and found that Research Policy and Social Studies of Science are the third and fourth most frequently referenced journals in articles published in Scientometrics.
Chen, McCain, White, and Lin (2002) drew upon citation and co-citation patterns derived from articles published in the journal Scientometrics (1981-2001).
Hou, Kretschmer, and Liu (2008) analyzed the structure of scientific collaboration networks in scientometrics at micro level (individuals) by using bibliographic data of all papers published in Scientometrics of the years 2002-2004.
Dutt, Garg, and Bali (2003) made an analysis of papers published by Scientometrics during 1978 to 2001 by scientometrics assessment on countries and themes distribution, comparison of institutions and co-authors.
The analysis of 816 papers published in Scientometrics during 2002-2008 showed that they were contributed by 57 countries (or regions). ... Most of the 57 countries were from Europe. Other major countries (or regions) had a larger number of papers were USA and Canada in North America, China, India, Taiwan, South Korea and Japan in Asia, Brasil in Latin America, and Australia.
Fig. 2 had clearly illustrated the average annual growth rates of TOP10 countries, from which we can conclude that USA, Belgium, Spain, China and Germany had higher growth rates and India had a negative growth rate.
From Fig. 3 we could see that all the TOP 10 countries had a higher number of times cited. It indicated that the papers contributed by those countries were of higher quality and had more impact.
In order to visualize the relative intensity of collaboration, this article introduced the concept of Relative Collaboration Intensity (RCI) indicator. The average number of collaboration countries (CC), average collaboration times (CT) and Relative Collaboration Intensity (RCI) of the 10 countries were given in formula (1), (2) and (3):
We found that Belgium had the highest relative collaboration intensity, and England, Netherlands and USA ranked 2, 3 and 4.
In order to make a clear vision about the collaborations among all the countries/regions (57), the country collaboration network had been made with the method of SNA by NWB. ... The largest connected component in the network had 37 nodes, and there is another small component with 2 nodes.
Fig. 4 showed the largest component with 37 countries, which depicted that Belgium, England and Hungary had formed an strong connection. The largest connection lied between Belgium and Hungary, and the collaboration times were 27. Fig. 4 also showed that although the USA had the largest number of papers, the collaboration activity was weaker than Belgium, Hungary, England and Finland. USA had paid much more attentions to collaborate with Canada, England and Australia. Netherlands had collaborated with many countries, however, the collaboration times were fewer compared to Belgium, England and Finland.
The data showed that Katholieke Univ Leuven (Belgium), Hungarian Academy Science (Hungary) and Leiden Univ (Netherland) ranked from first to third both in number of papers and times cited and all of them had a biggish advantage to others.
We select the Most-Cited paper (that had the highest value of times cited) of each TOP 10 TC/P institutions and get 10 Most-Cited papers finally. By analyzing their citing papers, we found that the citing papers which had cited the Most-Cited paper of each institution distributed mainly in the fields of information science & library science, computer science, information systems and interdisciplinary applications ....
So a conclusion could be made that although an institution did not have many papers or hold the advantage of research activities, it could carry out one or some significant works that had a great impact on the future development of information science & library science. And some research work on scientometrics had also affected the development of some other scientific fields, such as the work of Taipei Med Univ.
Another study carried out by Dutt et al. (2003) in scientometrics showed that the average number of authors per paper was 1.73 during the period of 1978-2001. We studied the average number of authors per paper published in Scientometrics 2002-2008 and found that the value was 2.29, which indicated that collaboration in scientometrics had been growing since 2001.
To analyze the intensity of co-authorship pattern, the whole data (816 papers) had been divided into two groups, which were single authored (271) and multi-authored (545). Compared to the result made by Dutt et al. (2003) that more than half of the papers were single authored, we found that the ratio of papers written by two or more authors had increased rapidly from 2002-2008.
Table 6 listed the TOP 10 authors according to their number of papers. Compared to Fig. 6 we could find that all the TOP 10 authors were appeared in the biggest collaboration cluster. It was interesting to note that the TOP 10 authors collaborated with each other either directly or indirectly.
Most of the TOP 20 cited references had distributed in big co-citation clusters shown in Fig. 7. ... All these four highly cited papers in the biggest cluster were focusing on the relationship between science and technology especially for the effect of science on technology. ... The second largest cluster contained 19 nodes, two of which were ranked in TOP 20. The topics were about indexes to quantify an individual's scientific research output (Hirsch, 2005). The third largest cluster included three nodes listed in TOP 20, whose topics were about author collaborations (Glanzel, 2001; Katz & Martin, 1997; Narin, Stevens, & Whitlow, 1991). There were another two clusters containing two TOP 20 nodes, and one had 10 nodes, whose topics were on co-citation networks (De Solla. Price, 1965;Small, 1973), the other had only two nodes published in Nature and Science individually with the topic of the scientific impact and wealth of nations (King, 2004; May, 1997).
The major topic were social network analysis (Wasserman & Faust, 1994), Matthew effect in science (Merton, 1968), author self-citation (Glanzel, Thijs, & Schlemmer, 2004), country research performance (Moed, 2002), evaluation indicators of publication and citation (Schubert & Braun, 1986) and the calculation of web impact factors (Ingwersen, 1998).
The analysis of 816 papers published in Scientometrics during 2002-2008 showed that they were contributed by 57 countries (or regions). ... Most of the 57 countries were from Europe. Other major countries (or regions) had a larger number of papers were USA and Canada in North America, China, India, Taiwan, South Korea and Japan in Asia, Brasil in Latin America, and Australia.
Fig. 2 had clearly illustrated the average annual growth rates of TOP10 countries, from which we can conclude that USA, Belgium, Spain, China and Germany had higher growth rates and India had a negative growth rate.
From Fig. 3 we could see that all the TOP 10 countries had a higher number of times cited. It indicated that the papers contributed by those countries were of higher quality and had more impact.
In order to visualize the relative intensity of collaboration, this article introduced the concept of Relative Collaboration Intensity (RCI) indicator. The average number of collaboration countries (CC), average collaboration times (CT) and Relative Collaboration Intensity (RCI) of the 10 countries were given in formula (1), (2) and (3):
We found that Belgium had the highest relative collaboration intensity, and England, Netherlands and USA ranked 2, 3 and 4.
In order to make a clear vision about the collaborations among all the countries/regions (57), the country collaboration network had been made with the method of SNA by NWB. ... The largest connected component in the network had 37 nodes, and there is another small component with 2 nodes.
Fig. 4 showed the largest component with 37 countries, which depicted that Belgium, England and Hungary had formed an strong connection. The largest connection lied between Belgium and Hungary, and the collaboration times were 27. Fig. 4 also showed that although the USA had the largest number of papers, the collaboration activity was weaker than Belgium, Hungary, England and Finland. USA had paid much more attentions to collaborate with Canada, England and Australia. Netherlands had collaborated with many countries, however, the collaboration times were fewer compared to Belgium, England and Finland.
The data showed that Katholieke Univ Leuven (Belgium), Hungarian Academy Science (Hungary) and Leiden Univ (Netherland) ranked from first to third both in number of papers and times cited and all of them had a biggish advantage to others.
We select the Most-Cited paper (that had the highest value of times cited) of each TOP 10 TC/P institutions and get 10 Most-Cited papers finally. By analyzing their citing papers, we found that the citing papers which had cited the Most-Cited paper of each institution distributed mainly in the fields of information science & library science, computer science, information systems and interdisciplinary applications ....
So a conclusion could be made that although an institution did not have many papers or hold the advantage of research activities, it could carry out one or some significant works that had a great impact on the future development of information science & library science. And some research work on scientometrics had also affected the development of some other scientific fields, such as the work of Taipei Med Univ.
Another study carried out by Dutt et al. (2003) in scientometrics showed that the average number of authors per paper was 1.73 during the period of 1978-2001. We studied the average number of authors per paper published in Scientometrics 2002-2008 and found that the value was 2.29, which indicated that collaboration in scientometrics had been growing since 2001.
To analyze the intensity of co-authorship pattern, the whole data (816 papers) had been divided into two groups, which were single authored (271) and multi-authored (545). Compared to the result made by Dutt et al. (2003) that more than half of the papers were single authored, we found that the ratio of papers written by two or more authors had increased rapidly from 2002-2008.
Table 6 listed the TOP 10 authors according to their number of papers. Compared to Fig. 6 we could find that all the TOP 10 authors were appeared in the biggest collaboration cluster. It was interesting to note that the TOP 10 authors collaborated with each other either directly or indirectly.
Most of the TOP 20 cited references had distributed in big co-citation clusters shown in Fig. 7. ... All these four highly cited papers in the biggest cluster were focusing on the relationship between science and technology especially for the effect of science on technology. ... The second largest cluster contained 19 nodes, two of which were ranked in TOP 20. The topics were about indexes to quantify an individual's scientific research output (Hirsch, 2005). The third largest cluster included three nodes listed in TOP 20, whose topics were about author collaborations (Glanzel, 2001; Katz & Martin, 1997; Narin, Stevens, & Whitlow, 1991). There were another two clusters containing two TOP 20 nodes, and one had 10 nodes, whose topics were on co-citation networks (De Solla. Price, 1965;Small, 1973), the other had only two nodes published in Nature and Science individually with the topic of the scientific impact and wealth of nations (King, 2004; May, 1997).
The major topic were social network analysis (Wasserman & Faust, 1994), Matthew effect in science (Merton, 1968), author self-citation (Glanzel, Thijs, & Schlemmer, 2004), country research performance (Moed, 2002), evaluation indicators of publication and citation (Schubert & Braun, 1986) and the calculation of web impact factors (Ingwersen, 1998).
沒有留言:
張貼留言