Category Archives: Networks

Emerging Methods and Tools for Sparking New Global Creative Networks – COINs15 Tokyo

Download “Emerging Methods and Tools for Sparking New Global Creative Networks” from the Proceedings of the 5th International Conference on Collaborative Innovation Networks (COINs15), Tokyo, Japan – Paper: [PDF] [arXiv] [arXiv PDF] Presentation: [PDF] Poster: [PDF]

Full text:

Proceedings of the 5th International Conference on Collaborative Innovation Networks (COINs15)

Emerging Methods and Tools for Sparking New Global Creative Networks

Jeff Horon, Elsevier Research Intelligence, 360 Park Ave S, New York, NY, 10010, USA


Emerging methods and tools are changing the ways participants in global creative networks become aware of each other and proceed to interact.  These methods and tools are beginning to influence the collaboration opportunities available to network participants.
Some web-based resources intended to spark new collaborations in creative networks have been plagued by dependence on fragmented or out-of-date information, having shallow recall (e.g. by being limited to a list of manually curated keywords), offering poor interconnectivity with other systems, and/or obtaining low end-user adoption.

Increased availability of information about creative network participants’ activities and outputs (such as completed sponsored research projects and published results, aggregated into global databases), coupled with advancement in information processing techniques like Natural Language Processing (NLP), enables new web-based technologies for discovering subject matter experts, facilities, and networks of current and potential collaborators.  Large-scale data resources and NLP allow modern versions of these tools to avoid the problems of having sparse/fragmented data and also provide for deep recall, sometimes within and across many disciplinary vocabularies.  These tools are known as “passive” technologies, from the perspective of the creative network participant, because the agent must undertake an action to use the information resources placed at his or her disposal.

Emerging “active” methods and tools utilize the same types of information and technologies, but actively intervene in the formation of the creative network by suggesting connections and arranging virtual or physical interactions.  Active approaches can achieve very high end-user adoption rates.

Both active and passive methods strive to use data-driven approaches to form better-than-chance awareness among networks of potential collaborators.  Modern instances of both types of systems generally support interconnectivity with other systems, and therefore expand the size of participants’ networks, resulting in a larger pool of potential collaborators from which to draw upon, within the system and additionally wherever the data is repurposed (e.g. into federated searches and customized applications).

Examples and Case Studies

“Passive” Networking Applications

The most widely deployed applications (providers) are: the Pure Experts portal (Elsevier), VIVO (DuraSpace), and Harvard Profiles (Harvard Medical School).  Each of these applications facilitates search and discovery of subject matter experts and their research activities and outputs.  These systems are generally organized and supported at the university level.  These applications are also federated into multi-institutional search frameworks including Direct2Experts and CTSAsearch – both of which are open to all three of the networking applications above, as well as other less widely deployed applications.

“Active” Networking Applications

Efforts toward active networking interventions are sometimes made with ‘researcher speed dating’ activities, but these generally rely on an audience with some mutual interests being gathered together (e.g. at a conference or symposium) and pairings are typically random.  Despite the existence of predictive factors for propensity to collaborate and likelihood of achieving team goals (e.g. obtaining external funding for research projects)[i], data-driven active networking methods are comparatively rarely used.  Prior case studies in active networking include:

Team design for large center and team science proposals

The University of Michigan Medical School assisted a principal investigator applicant for a large center grant with team formation, based on identifying potential participants publishing or having sponsored projects in subject matter related to the center.  This allowed for discovery of related expertise by analyzing term co-occurrence, and then discovery of the subject matter experts working with those concepts.  Multiple rounds of iteration resulted in a list of keywords, stemmed to related key terms, such that the list was both inclusive of the desired family of concepts and exclusive of ‘false positive’ matches.

Suggested casual interactions at a physical event

At an institute launch event, the University of Michigan employed search methods similar to those above for objective detection of researchers working in related topic areas, to supplement institute founders’ knowledge of researchers working in relevant topic areas with information about previously-unknown researchers also working in these topic areas.  Objective detection allowed for increased inclusiveness and comprehensiveness of the launch conference invitee list.

Launch event organizers solicited survey responses from participants concerning areas of methodological expertise, methodological needs for upcoming projects, and areas of interest within several pre-identified areas related to the institute.

Attendees were matched based upon expressing strong mutual interest in a topic and/or by study method, in situations where one researcher expressed a need for expertise in a method and another research expressed the ability to share methodological expertise in the same method.  Reciprocal methodological need/provision matches were considered especially strong matches (Figure 1):


Figure 1:   A generalized example of an especially strong match

Existing collaboration data covering co-authored publications and co-participation on sponsored projects were used to rule out matches who had collaborated in the past.

To maximize the chances strong matches would interact, the seating chart was also arranged to place strong matches at the same tables.  This event also included conversation-provoking material, including a visualization of attendees arranged in a social networking diagram according to indicated areas of strong interest.

The matching process proved to be very flexible and was used to support a novel approach to bridging mentorship gaps in pediatric research[ii].

Scheduled interactions at a physical event

The University of Texas System M.D. Anderson Cancer Center has in recent years built into a key global cancer conference activities for scheduled networking interactions.  The survey mechanism is similar to the University of Michigan example above, as are the recommendations, but there is also accommodation for arranging meetings including generally a mix of online meeting coordination, dedicated meeting time available, and dedicated meeting spaces available.  Given rotating global locations and varied attendees from year-to-year, priority is given to matches from different institutions as there may only be one time they are physically co-located.

In addition to the meetings booked during a specific speed dating event window in the conference program, the project team also noted a number of off-hours and informal meetings taking place, driven in part by the recommended matches.


These emerging methods and tools suggest the existence of repeatable strategies for facilitating data-driven matching and better-than-chance interactions designed to spark new global creative networks.  As these methods become further systematized and see wider adoption, they are poised to influence larger numbers of creative networks and their participants.

[i] Lungeanu, A., Huang, Y., and Contractor, N.S. (2014) “Understanding the assembly of interdisciplinary teams and its impact on performance.” Journal of Informetrics.  8(1):59-70.

[ii] Nigrovic, P.A., Muscal, E., Riebschleger, M., et. al. (2014) “AMIGO: A Novel Approach to the Mentorship Gap in Pediatric Rheumatology” Journal of Pediatrics 164(2):226-7.e1-3.

Places & Spaces: Mapping Science Exhibit

The Places & Spaces: Mapping Science Exhibit visited the University of Michigan Hatcher Graduate Library from March 7 through May 24, 2011.

From my trip to the exhibit — the Network Science Research poster from University of Michigan contributors featured prominently at the entrance:


 From the poster:

The works presented here were invited to accompany the international Places & Spaces: Mapping Science exhibit on display at the University of Michigan Library in March 1 to May 24, 2011.

The maps showcase excellence in network science research and education at the University of Michigan. They were created in many different areas of science, including medicine, bioinformatics, sociology, organizational studies, information science, and physics, to advance our understanding of the structure and dynamics of networks.

On a practical side, Jeff Horon’s map demonstrates the innovative usage of science analysis and mapping in support of team formation and grant proposal writing that helped secure a P30 award by National Institutes of Health that finances a new Nutrition Obesity Research Center at the University of Michigan.

It is our hope that the Mapping Science exhibit will inspire future collaborations across disciplinary and geospatial boundaries, the application of effective visualization techniques to communicate the results of these studies, and the adoption of advanced data analysis and visualization methods to improve daily decision making.

Download the poster: PDF

The exhibit features maps describing the power of maps, reference systems, and forecasts as well as science maps for economic decision makers, policy makers, and scholars. These are the ‘stars of the show’ but I won’t go into detail here since they are already well-described on the exhibit website.

The exhibit also features WorldProcessor Globes reflecting ‘Patterns of Patents & Zones of Invention’ over time and across geography:


Patterns of Patents & Zones of Invention [WorldProcessor #286]

“This globe plots the total amount of patents granted worldwide, beginning in 1883 with just under 50,000, hitting 650,000 in 1993 (near the North Pole), and (shifting the scale to the southern hemisphere) continuing to 2002 on a rapid climb towards 1 million. Geographic regions where countries offer environments conducive to fostering innovation are represented by topology. Additionally, nations where residents are granted an average of 500 or more US patents per year are called out in red by their respective averages in the years after 2000. © 2005 Ingo Gunther”


And the ‘Shape of Science’ given “quantified connectivities and relative flows of inquiry within the world of science”:


“This rendering is of a prospective tangible sculpture of the Shape of Science, based on the research of Richard Klavans and Kevin Boyack, spatializing the quantified connectivities and relative flows of inquiry within the world of science. © 2006 Ingo Gunther w/ Stephen Oh”


The exhibit also includes interactive video monitors and hands-on maps for kids.

 Attributions and Acknowledgements:

Network Science Research Poster

“Several University of Michigan faculty created maps included in the exhibit: Santiago Schnell, Molecular and Integrative Physiology; Lada Adamic, School of Information; M. E. J. Newman, Physics; Jeff Horon, Medical School; Helena Buhr, Natalie Cotton, and Jason Owen-Smith, Sociology and Organizational Studies.”


Overall Exhibit

Places & Spaces is curated Dr. Katy Börner and Michael J. Stamper, School of Library and Information Science at Indiana University. Places & Spaces also receives input from the Places and Spaces Advisory Board. The exhibit is sponsored by the National Science Foundation under Grant No. IIS-0238261, CHE-0524661, IIS-0737783 and IIS-0715303; the James S. McDonnell Foundation; Thomson Scientific/Reuters; Elsevier; the Cyberinfrastructure for Network Science Center, University Information Technology Services, and the School of Library and Information Science, all three at Indiana University. Some of the data used to generate the science maps is from Thomson Scientific/Reuters.

Thank you to Dr. Katy Börner and Michael J. Stamper for curating the exhibit. Thank you to Tim Utter and Rebecca Hill for bringing the exhibit to the University of Michigan.

This exhibit was mentioned in press at:

Heritage Newspapers Online and Print Editions

Approachable Network Analysis

Materials from my presentation Approachable Network Analysis are now available: PDF

From the University of Michigan Business Intelligence Community of Experts site:

Approachable Network Analysis


Jeff Horon
Business Analyst
Medical School Grant Review & Analysis Office


Unlock the power of network structures in your data. Learn how to build and analyze networks to gain insights through relationship analysis. Apply approachable techniques and free, user-friendly software. Transform the data you have into the data you need – from relational databases and unstructured text to common network structures.

Jeff detailed his work in the Medical School Grant Review & Analysis Office. Examples will include: Identifying networks of collaborators from eResearch Proposal Management [eRPM PAF] data, discovering networks of concepts in unstructured text, and use cases from other administrative data sets. Jeff’s presentation included:

-“Networks 101″ – The basic building blocks of networks
-How people in any campus unit can apply network analysis
-An emphasis on approachable techniques and free, user-friendly software
-Strategies for effectively visualizing and sharing network-driven insights
-Tools, tips, and tricks

Presented examples were produced with NodeXL, which is a product of Microsoft Research and university collaborators. NodeXL is very approachable — you can produce a network visualization by simply copying and pasting in two columns of properly-formatted data (i.e. an ‘edge list’). It computes a desirable set of network metrics, including eigenvector centrality. Layout, visual encoding, and aesthetics are top notch. It installs into Excel 2007+ as a template.