Collaboration with Domain Experts Amitabh Varshney University of - PowerPoint PPT Presentation

Collaboration with Domain Experts Amitabh Varshney University of Maryland

Overall Research Funding Trend Declining/Constant Federal Research $$ Science, March 2017 AAAS, June 2016

Research Funding by Discipline AAAS Sept 2017

Interdisciplinary Research

David Fushman Selected Collaborations Chem & BioChem Bill Dorland Physics Jon Dinman CBMG Sergei Sukharev Biology Eugenia Kalnay AOSC Lee Mundy Astronomy

Selected Collaborations Lisa Shulman Joseph JaJa Pam Clark Neurology Computer Engg Public Health Kris Marsh, Sarah Murthi, Mario Dagenais, Rashawn Ray Tom Scalea Optics Sociology Shock Trauma

Domain Expert Collaboration • Selection of Problem • Selection of Domain Expert • Approach to collaborative problem solving • Cost of collaboration • Measures of success

Domain Expert Collaboration • Selection of Domain • Selection of Domain Expert – Selection of Problem

Domain Selection Personal interest, advice from mentors, potential for impact • Why now? • – What is the catalyst • Why here? – Institution, Location • Why this ? – Work on several things, but reserve deep dives for select few – At most 8 bullets!

Domain Expert Selection • Career stage match preferable • Taste in problems: elegance, potential impact – Past results are a prediction of future returns – What are important problems in your field? What are you working on? • Matching goals, personality • Funding balance – independence or mutual dependence

Approach to Collaboration • Driving Application Approach – Enhancing the collaborator’s application will enrich my approach to Vis Working on real problems, which if solved, will have impact • • Working on full problem (not a self-selected subset) • Planning and strategy 1:1 with Domain Expert – Cannot delegate to students – Cannot substitute by a committee

Costs of Collaboration • Limit, but do not eliminate, the technical service component • Challenge for students – need basic knowledge of the domain • Falling through the cracks of interdisciplinary collaboration • Challenges in publication – We focus on novelty of technique – We usually discount novelty of the domain

Evaluation • Does the Domain Expert use it in their workflow? – Work with the legacy workflow • Broad, statistically significant, user studies are difficult – Few experts available to study, typically co-authors – Design decisions, lessons learned

Sources • You and your research, Richard Hamming, June 1995 • Computer Scientist as a Toolsmith II, Fred Brooks Jr., CACM, March 1996 • What is the key best practice for collaborating with a computational biologist, A. Carpenter, Cell Systems, July 2016