Hacking the Web of Science data? From bibliometric projects to researcher portals

Transcription

1 Hacking the Web of Science data? From bibliometric projects to researcher portals ALA June 2017 Moderator- Ann Beynon Panelists: Eamon Duede Junming Huang Anita Foster Gene Springs

2 Web of Science Core Collection: a single source for over 65 million publication records Comprehensive coverage Over 65 millions bibliographic records 18,000 journals, 180,000 conference, 80,000 books 2 SCIENCE CITATION INDEX EXPANDED 1900-present 48.4 million BOOK CITATION INDEX 2005-present 1.1 million ARTS & HUMANITIES CITATION INDEX 1975-present 4.7 million SOCIAL SCIENCE CITATION INDEX 1900-present 8.6 million EMERGING SOURCES CITATION INDEX 2015-present (2005 coming 2017) 374,000 CONFERENCE PROCEEDINGS CITATION INDEX 1990-present 9.6 million Massive citation network Explore over 1 billion Cited References, back to 1900, to form an extensive network of scientific connections. XML format Data delivered in XML format to enable large-scale analysis, integration with other datasets, and use in statistical and visualization programs. Historical analysis Data start in 1900, enabling long-term trend analysis. Complete author data Conduct extensive analysis of researchers and their collaborations with indexing of all author names, along with ResearcherID and ORCID identifiers, and algorithmic author disambiguation. Complete address information Complete indexing of all author addresses enables analysis of institutions, cities, states, countries, and regions. Address unification. Funding acknowledgements Connect research to its funding sources from around the world and across sectors.

3 HOW CAN YOU ACCESS WEB OF SCIENCE DATA? 3 Online API FACULTY ACTIVITY REPORTING SYSTEMS DATA WAREHOUSES RESEARCH PORTALS Data delivery

4 APIS ENHANCE INSTITUTIONAL SYSTEMS WITH TRUSTED ARTICLE METADATA AND CITATION METRICS 4 INSTITUTIONAL REPOSITORIES RESEARCH PORTALS DATA WAREHOUSES FACULTY ACTIVITY REPORTING AND RESAERCH INFORMATION MANAGEMENT SYSTEMS

5 APIS for data integration: Institutional Repository 5 Queensland Univ Technology: Utilizing Article Match Retrieval API, institutions can place citation counts along with link to Web of Science Core Collection Source: Queensland University of Technology:

6 APIS for data integration: VIVO Univ Florida VIVO: Created in 2009 and features over 50,000 publications using Web of Science Core Collection API 6 Source: University of Florida VIVO:

7 APIS for data integration: VIVO 7 Univ Idaho VIVO: Created in 2012 and features over 14,000 publications using Web of Science Core Collection API Navigate from VIVO Implementation to Web of Science Core Collection to view full metadata Source: University of Idaho VIVO:

8 Web of Science Core Collection raw data for research

9 Eamon Duede Executive Director Knowledge Lab Computation Institute Committee on Conceptual and Historical Studies of Science University of Chicago

10 Research center at the heart of the University of Chicago s Computation institute 7 University of Chicago Affiliated Faculty: Computer Science, Sociology, English Literature, Organizations and Markets, Genomics 6 Postdoctoral Scholars 3 Developers and Data Scientists 10+ Graduate and Undergraduate Research Assistants

11 Research center at the heart of a nationally distributed network of researchers 40+ Scientists and Scholars 13 Premier Institutions

12 Science as a System

13 The Scientific system is a heterogeneous assemblage of things that are interesting on their own and (sometimes) constitute knowledge when combined in certain ways

14 The Scientific system is a heterogeneous assemblage of things that are interesting on their own and (sometimes) constitute knowledge when combined in certain ways Things like

15 Authors Institutions Questions / Problems Methods

16 And interesting stuff like: Chemicals Patents Claims Genes Theorems Theories

17

18

19

20

21

22

23

24 Researcher Interest Comparisons the K nearest faculty members to a given researcher

25 Researcher Opportunity Alignments the K nearest faculty members to a given Funding Opportunity

26 3 Shades of Data Public Publically accessible Not sensitive No protection required Eg. W iki, Google scholar Confidential Private data Limited disclosure risk Legal/ contract ual obligations Eg. W OS, JSTOR Regulated Sensitive data Requires disclosure control Govt. regulated Eg. AURA db, PHI dat a, SSN etc

27 Cloud Kotta Cloud Kotta: Secure and Scalable Data Analytics in the Cloud. IEEE BigData, 2016 A Secure Data Enclave and Analytics Platform for Social Scientists. IEEE escience, 2016 Enabling Interactive Analytics of Secure Data in Cloud Kotta Science Cloud, 2017

28 With private data-sets comes great responsibility A significant fraction of the 10TB we EBS manage is sensitive/proprietary data: EC2 Web of Science - from Thomson Reuters (1TB) We want to make this data accessible to our colleagues and collaborators, but secured within our infrastructure.

29 Cloud Kotta Cloud Kotta is a cloud-based platform that enables secure and cost-effective management and analysis of large, potentially sensitive data Data is stored securely and migrated between storage tiers depending on access patterns and pre-defined policies The platform automatically provisions cloud infrastructure to host user submitted jobs Role based access model for security in Malayalam Kotta means Fortress * Pictured: Mehrangarh Fort at Jodhpur, Rajasthan

30 Lots of user Interfaces Jupyter notebooks Web Interface REST API Command Line Interface

31 Automated Storage Management

32 Elastic Provisioning

33 Security Model Principle of least privilege throughout Log in with Amazon Users are assigned roles Policies permit access to resources for individual roles Instances are granted a trusted role that allows them to switch to a user role temporarily in order to inherit user permissions (e.g., access secure data) Compute layer is hosted within a private subnet enclosed within a VPC

34 Stewardship

35 Full Job Management and Publication

36 Full Job Management and Publication

37 Exploring Science of Science with Web of Science dataset 24 Jun 2017 Junming Huang Northeastern University

38 Web of Science dataset 55 MILLION science and social science papers. 98 MILLION authorships (author-paper pairs). 181,453,658 collaboration links between authors. 759 MILLION citation links between papers. 679,358,136 if self-citations removed All science and social science journals during categories, including Arts & Humanities (14), Science & Technology (116), Social Sciences (24).

39 Accessing Web of Science dataset Raw meta data Structured data Title Author Journal Keywords Affiliations

40 Accessing Web of Science dataset Structured data Store & Index Efficiently accessible data Table Author Author Author Table Address Address Address Table Paper Paper Paper Table Citation Citation Citation Table Journal Journal Journal

41

42

43

44 Accessing Web of Science Filter and retrieve millions of records within minutes

45 Accessing Web of Science dataset University server Authorized researchers

46 Accessing Web of Science dataset Zoom Six top ranked countries for total number of scientists Czech Republic Japan (the most imbalanced country) Romania (the most balanced country) Authorized researchers

47 Physics Anatomy: What is physics? Web of Science, 27M papers???? Not Physics? P?? P? Interdisc. Physics 3.2M papers P?? P? P?? P IP Citing Physics 1.3M papers Core Physics 2.4M papers 242 journals Physics-cited 0.6M papers Defining physics in a recursive way: Physics is what physicists do. Sinatra, Deville, Szell, Wang, Barabási, Nature Physics, 11, 791 (2015)

48 Physics Anatomy: How physics evolves in century? Structure small fields are more selfreferential and insular Flow more knowledge from core to interdisciplinary Sinatra, Deville, Szell, Wang, Barabási, Nature Physics, 11, 791 (2015)

49 Wang, Song, Barabási. Quantifying Long-Term Scientific Impact. Science 342(6154). (2013) Citation: Decoding the dynamics Describing a thousand of curves with one pattern

50 Wang, Song, Barabási. Quantifying Long-Term Scientific Impact. Science 342(6154). (2013) Citation: Decoding the dynamics Hidden parameters characterizing journals Predicting citations of a single paper

51 Credit: Allocating among co-authors Which author in a paper should get the most credit? Inferring credit allocation in a collective way Shen, Barabási, Collective credit allocation in science. Proceedings of the National Academy of Sciences 111(34). (2014)

52 Credit: Allocating among co-authors Accurately predicting who should win Nobel Prize Young scientists cumulate credits Shen, Barabási, Collective credit allocation in science. Proceedings of the National Academy of Sciences 111(34). (2014)

53 Career: Exploring gender gap in academia Zoom Publications a b c Underrepresented female scientists: population, publication and citation Six top ranked countries for total number of scientists Czech Republic Impact d e Japan (the most imbalanced country) Romania (the most balanced country) f Why? Huang, Sinatra, Barabási. In preparation.

54 Career: Exploring gender gap in academia a b c d e Gender gap will be significantly reduced once we control the excessive dropout rate of female Huang, Sinatra, Barabási. In preparation.

55 Citation: Bias to correct? Researchers in some countries get excessive domestic citations Yifang Ma, Roberta Sinatra, Michael Szell, Albert-László Barabási. In preparation

56 Thank you Kim Albrecht Pierre Deville Junming Huang Yasamin Khorramzadeh Yifang Ma Roberta Sinatra Chaoming Song Michael Szell Dashun Wang László Barabási

57 Web of Science Dataset at an Academic Library Anita Foster, Electronic Resources Officer Gene Springs, Collections Strategist The Ohio State University Libraries ALA Annual June 24, 2017

58 Campus & Data Translational Data Discovery Theme in 2014 Launch of Translational Data Analytics Program in 2015 Interdisciplinarity across campus Research Commons space envisioned

59 Research Commons Services for Research Commons begin in Summer 2014 Physical space opens in January 2016 Collaborative space for graduate students and faculty Computer lab with statistical data software Data visualization lab Expansion of services, campus launch Became the point of need space for data services

60 Strategic Collection Areas of FY17 3 Collection Areas prioritized for one-time funds in FY17 Streaming Media (continuation from FY16) Business Resources DATA Pilot Programs

61 FY17 Data Pilot Parameters none! Budget no specifics! Process & workflow we ll learn! Multi-year, pay in advance agreements preferred! Communication inclusive!

62 Research Data in the Library Task Force Long-gestating task force focused on research data services charged and deployed in June 2016 Charge: Conduct an environmental scan to reflect current OSUL services and practices that support research data Include a statement on the challenges and complexities which currently prevent us from offering support Recommend solutions and next steps for delivering a robust suite of services around research data Representation from: Research Commons, Subject Librarians, Archives, Acquisitions, Institutional Repository, Health Sciences Library, (Collection Strategy)

63 Other Factors New Resource Proposal Process Submitted Research Data in the Library Task Force report (11/16) Re-envisioned Research Data in the Library Task Force (1/17) Re-submitted Research Data in the Library Task Force report (4/17) Creation of Data Working Group (5/17)

64 Web of Science Patron Subject Librarian A Subject Librarian B Collection Strategy/Electronic Resources Librarians Subject Library A/B Research Commons Staff Data Management Services Librarian Interim Head of Acquisitions Electronic Resources Officer Data Management Services Librarian Patron Need met???

65 Web of Science Purchased Web of Science Core Collection Data Set + 3 years of data refreshes Initial Library Contact/Communication: November 2016 Statement of Work Agreement Fall Semester 2017

66 References Swanson, J., & Rinehart, A. K. (January 01, 2016). Data in context: Using case studies to generate a common understanding of data in academic libraries. The Journal of Academic Librarianship, 42, 1, ; Research Commons: Translational Data Analytics Institute:

67 For more information Anita Foster Gene Springs

68 Thank you Ann Beynon