Upload
mindtrek
View
14
Download
1
Embed Size (px)
Citation preview
Visualizing Co-authorship Networks for Actionable Insights: Action Design Research Experiment
Jukka HuhtamäkiTampere University of Technology
Academic Mindtrek 2016@jnkka #cobweb #weakties
Context
01.05.2023 2http://bit.ly/cobwebtut
• Computational methods for intelligent matchmaking for knowledge work
• Increase serendipity and inter-disciplinary cross-pollination of ideas through data-driven matchmaking
• Tampere University of Technology: IHTE, NOVI, and Mathematics• Funded by the Academy of Finland
Objective of the experiment• How to use visual network analytics to create
additional value for bibliographical data? • More specifically, what kinds of insights does
network analytics of bibliographic data afford?
• How should the digital research infrastructure (academic ecosystem) support data-driven analytics?
01.05.2023 3
• Current Research Information Systems (CRIS) are used to collect, manage, and search bibliographic data
• The CRIS used in the experiment must be operated manually with a user interface
• One can export a set of research results only if there are less than 1000 articles found
• The exported data includes only a fraction of the data maintained by the CRIS
Minimum experimentable product
Information system includes• social,• information, &• technologyartifacts
(Lee, Thomas & Baskerville, 2015; Vartiainen and Tuunanen, 2016)
Social artifact1. “Start with what you know, then grow” (Heer
and boyd, 2005)2. What are the mechanisms behind the structure:
organizational, research groups, shared interests?
3. Explore and benchmark ways of working: What lead to productivity and success? International collaboration?
4. Networks provide a way to measure research in a more systemic manner
01.05.2023 6
Information artifact• Two key entities: individual articles and
collections or articles1. No unique identifiers for authors in article
metadata 2. No organizational information on the authors
available3. Cap of 1000 articles to be exported4. No explicit license for the data (Elsevier Pure)
01.05.2023 7
Technology artifact• Data access and processing in Python
(Pandas and NetworkX)• Network analysis and visualization in Gephi• Interactive network provision in Gexf.js
• Developing a visual analytics system operating in self-service mode could be conducted as an open source project
01.05.2023 8
Discussion – please join! • Use Open data license (CC) for bibliographical data with a specific
open data license• Provide a REST interface for accessing bibliographic data • Provide a REST interface for the search• Provide complete data on articles and other entities in JSON and
XML• Provide unique identifiers for articles, authors, and organizational
and other entities• Apply linked data practices to support traversing the metadata (URI
schema)• Do not limit the number of articles that can be fetched from the
repository. • To enable the launch of external, third-party visual analytics tools
(see the paper and Salonen and Huhtamäki (2010) for details)01.05.2023 9
Concluding remarks• We need a digital ecosystem for research• Open computational access to bibliographic
data is an important first step• In national level, VIRTA REST gives some
support to data access (delay, not tested)• “we conclude with the strongest possible
recommendation for university policy makers to step into the open science sphere”-open bibliographic data is the imperative first step
01.05.2023 10
Thank you!
01.05.2023 11
Jukka Huhtamäki
fi.linkedin.com/in/jukkahuhtamaki/