Tag Archives: SciTS

Data-Enabled and Spontaneous Researcher Networking at an International Conference – SciTS 2014

Download “Data-Enabled and Spontaneous Researcher Networking at an International Conference” from the Science of Team Science [SciTS] 2014 conference: PDF

Abstract:

This case study explores the use of both data-enabled and spontaneous researcher networking activities among attendees at an international conference based in Seoul, South Korea, in an attempt to utilize knowledge from the Science of Team Science field and discern best practices in their application.

Data-enabled networking activities included an advance survey of all attendees to find participants willing to participate in networking activities, and then suggesting networking partners based upon methodological expertise, methodological needs, and common interests. Response rates and participant feedback will be discussed.

Spontaneous networking activities were also made available to attendees stopping by a physical networking space made available for the duration of the conference. Activities included live browsing of research networking tools and a system for making requests related to research and responding to the requests of others, based upon sociological theories of reciprocity. Participation rates and participant feedback will be discussed.

Data-Driven Matching for Fostering Relationships among Scientists: Quantitative Assistance for a Human-Centered Process – Science of Team Science 2012

Download our poster: PDF

Authors

Jeff Horon, MBA, Elsevier, Inc.
Meredith Riebschleger, MD, University of Michigan Mott Children’s Hospital

SciTS Topic

Other (Methods for Team Formation / Researcher Networking)

Key Words

Team Formation, Researcher Networking, Data-Driven Networking, Interventional Networking, Algorithm-Based Matching, Dyads, Mentoring, Mentor-Mentee

Abstract

Data-driven matching of scientists for the purpose of fostering working relationships has the potential to improve and expedite the matching process. The AMIGO Steering Committee, which matches Pediatric Rheumatology mentors and mentees nationally, employed survey data and a matching algorithm to expedite matching of mentor-mentee dyads. In pilot testing, the group matched 20 dyads from pools of 20 mentees and 49 mentors, completing the process in only 1 hour. Early indicators are that these matches were well-formed. The project team continues to track the experimental (algorithm-assisted pilot) group against the control (human-only matching process) group. These findings invite further study by team science researchers working to predict successful team formation in the matching of dyads and larger groups of scientists. These findings also suggest new best practices for practitioners of scientist matching.

Typically, such a third party matching is accomplished using human judgment about an implicitly calculated probability for success, but in situations where many matches are required, the complexity of the problem quickly scales beyond human limits like available time and memory capacity.

Using data-driven matching via algorithms can provide meaningful assistance. Algorithms can account for more persons and more facts per person during the matching process and can evaluate a significantly higher number of potential matches. Importantly, data-driven matching is not mutually exclusive with human-driven matching. The AMIGO Steering Committee used the output of the data-driven match process as an input to the human-driven match process.

Beyond the remarkable time savings involved in matching 20 dyads in an hour, preliminary results reflecting match quality are positive. Of these dyads, 9 met in person at an event following the match process. As of one month after the initial match (with 20 of 20 dyads responding), 19 of the dyads had made initial contact, 13 had exchanged CVs, and 15 had spent at least 30 minutes in discussion of the mentee’s career. One dyad had to be re-matched for reasons not captured in the questionnaire.

The project team anticipates expanding the process to more mentor-mentee dyads in the future and adjusting the algorithm based upon further analysis of how the results of the human matching process differed from the algorithm matching process.

Further study is suggested for the formation of larger groups.