Börner, K., Huang, W., Linnemeier, M., Duhon, R. J., Phillips, P., Ma, N., ... & Price, M. A. (2010). Rete-netzwerk-red: analyzing and visualizing scholarly networks using the Network Workbench Tool. Scientometrics, 83(3), 863-876.

Börner, K., Huang, W., Linnemeier, M., Duhon, R. J., Phillips, P., Ma, N., ... & Price, M. A. (2010). Rete-netzwerk-red: analyzing and visualizing scholarly networks using the Network Workbench Tool. Scientometrics, 83(3), 863-876.

information visualization

計算科學學(computational science)的眾多研究人員和實務工作者對於資料、演算法、大量硬碟空間與計算能力的需求持續地成長，為了因應這個現象有必要提供一個電腦網路基礎架構(cyberinfrastructure)來支援進階的資料獲取、儲存、管理、整合、視覺化與分析。本研究提出兩個這方面的實務做法：一為一個能夠提供論文、專利和獎補助計畫等各種計算科學學資源的資料庫(the Scholarly Database, SDB)，另一則為一個能夠提供網絡分析、建立模型和視覺化的軟體工具(the Network Workbench, NWB)。SDB(http://sdb.slis.indiana.edu)包括Medline論文、USPTO專利、NSF及NIH獎補助計畫，超過2300萬筆紀錄，提供作者、題名、摘要和全文等欄位的搜尋，也可以限定年份的範圍。NWB(http://nwb.slis.indiana.edu)可以接受NWB、GraphML、Pajek.net、Pajek.matrix、XGMML、TreeML、CSV等七種檔案格式，並且提供格式轉換的功能；在提供科學計量學的應用方面，NWB也可以接受SDB、Google Scholar、Thomson Reuters ISI Scientific、Scopus、NSF獎補助資料庫的輸出，以及 EndNote和BibTeX等書目管理軟體的格式。NWB不僅能夠做為科學計量學的工具，同時也可以提供物理、生物醫學和社會科學的分析用途，因此這套軟體工具能夠執行網絡分析，產生網絡模型並加以驗證，同時也有視覺化工具以便與網絡互動來增進探索與了解。因此，此一工具的另一項優點是能夠提供跨學術領域的資料集與演算法。本研究並且以三個實驗來說明這兩個實務做法的應用。首先是美國主要大學獲得NSF獎補助的情形分析，並且從這些獎補助計畫發現共同研究者(co-investigator)的合作網絡以及網絡中主要的研究者。其次是以網絡科學(network science)此一科際整合的學科為例，分析來自Eugene Garfield、Stanley Wasserman、Alessandro Vespignani和 Albert-La´szlo´ Baraba´si等不同領域但對此學科有重要貢獻的科學家的合作網絡(collaboration networks)以及著作、引用和H指標(H-index)的不同。最後一個實驗案例是針對某一個正在興起的領域，例如RNA干擾(RNA interference, RNAi)，探討是何種主題大量出現，也就是哪一個主題在什麼時候開始活躍的問題。

Most studies use either Thomson Reuters’ Web of Science (hereafter WoS) or Elsevier’s Scopus, as they each constitute a multi-disciplinary, objective, internally consistent publication database. A number of recent studies have examined and compared the coverage of WoS, Scopus, Ulrich’s Directory, and Google Scholar (hereafter GS) (Meho and Yang 2007; Pauly and Stergiou 2005).

Cyberinfrastructures, i.e., the programs, algorithms and computational resources required to support advanced data acquisition, storage, management, integration, visualization and analysis (Atkins et al. 2003), address the ever-growing need to connect researchers and practitioners to the data, algorithms, massive disk space and computing power that many computational sciences require (Emmott et al. 2006).

The Scholarly Database (hereafter SDB) at Indiana University aims to serve researchers and practitioners interested in the analysis, modelling, and visualization of large-scale scholarly datasets. ... The online interface at http://sdb.slis.indiana.edu provides access to four datasets: Medline papers, U.S. Patent and Trademark Office patents (USPTO), National Science Foundation (NSF) funding, and National Institutes of Health (NIH) funding—over 23 million records in total, see Table 1.

SDB supports search across paper, patent, and funding databases. To initiate a search, enter the search term(s) into creators (author/awardee/inventor), title, abstract, and full text (keywords and other text) fields, select a year range and database(s)...

The Network Workbench (NWB) Tool (http://nwb.slis.indiana.edu) is a network analysis, modelling, and visualization toolkit for physics, biomedical, and social science research (Herr et al. 2007). The basic interface comprises a ‘Console’, ‘Data Manager’, and ‘Scheduler’ Window as shown in Fig. 2. The top menu provides easy access to relevant ‘Preprocessing’, ‘Modeling’, ‘Analysis’, ‘Visualization’, and ‘Scientometrics’ algorithms.

Users of the NWB tool can
• Load sample datasets or their own networks and formats.
• Perform network analysis with some of the most effective algorithms available.
• Generate, run, and validate network models.
• Use different visualizations to interactively explore and understand specific networks.
• Share datasets and algorithms across scientific boundaries.

The loading, processing, and saving of seven file formats (NWB, GraphML, Pajek.net, Pajek.matrix, XGMML, TreeML, CSV) and an automatic conversion service among those formats is supported. Relevant for science of science studies, the NWB Tool can read data downloaded from SDB, Google Scholar, Thomson Reuters ISI Scientific, Scopus, and the NSF award database as well as EndNote and BibTeX formatted data.

The first study aims to answer: What active funding portfolios do major U.S. universities have, what scholarly co-investigator networks does this funding inspire/support, and what roles do investigators play, e.g., gatekeeper, using betweenness centrality measures (BC) (Freeman 1977), number of collaborators via node degree, total funding amount?

There are interesting differences in the funding portfolios of these universities. Michigan has clearly the largest number of currently active NSF awards totalling 497. With $546 million, Cornell has the highest total award amount. Cornell also has the largest giant  component with 67 investigator nodes and 143 collaboration links, indicating much crossfertilization across different disciplines. Cornell also happens to employ the investigator who currently has the highest total award amount: Maury Tigner. Note that being the main investigator on one major center grant and several campus equipment grants can easily result in multi-millions to spend over many years.

A closer examination of the largest component of the Cornell co-investigator network shown in Fig. 3b, reveals that Steven Strogatz has the highest betweenness centrality (BC), effectively bridging between several disciplines, and Daniel Huttenlocher has the highest degree, i.e., the most collaborations with others in this network.

The second study addresses the questions: Do researchers which come from different domains of science but make major contributions to one and the same domain, e.g., network science, grow different collaboration networks? How much do their publication, citation, and H-index dynamics differ?

Exemplarily, four major network science researchers were selected: Eugene Garfield and Stanley Wasserman, Alessandro Vespignani and Albert-La´szlo´ Baraba´si of the Network Workbench project: Data for all four male researchers was downloaded from Thomson Reuters in December 2007.

While Baraba´si’s and Vespignani’s co-author networks are strongly interlinked with Stanley and Vazquez as major connectors with high betweenness centrality values, Garfield’s and Wasserman’s networks are unconnected to the networks of any of the three other researchers.

The third study aims to answer: What topic bursts exist in an emerging area of research, e.g., RNA interference (RNAi) research? Exactly what topics are active and when?