ASIST2010: Structure and evolution of scientific collaboration networks in a modern research collaboratory

Alberto Pepe, UCLA (dissertation award winner)

🙂 hasn’t looked at his dissertation for 4 months so he might be rusty, ask his advisor

look at collaboration in a collaboratory where work is interdiscplinary and distributed. Physical and virtual spaces… At CENS, sensor network research, about 300 researchers, multidisciplinary (ee, cs, stats, bio, environ sci, urban planners, sociology, media). Multi-sited within southern California. He was a participant observer.

Important to study relations instead of things bcs of connectedness of sci collabs. Looked at co-authorship, communication (mailing lists), acquaintanceship

What is the topology of the network, what is the structure of CENS and how has it evolved, how are the networks related to each other – what is the role of communication and acquaintanceship on co-authorship?

Authorship data – CENS annual reports are the official listing of all of the publications from the collaboration (lucky this exists – doesn’t for many orgs). 600 publications over 7 annual reports. 400 conf papers. Yr 2 was most productive, dip yr 5 then rise. Most papers have 2-3 authors. Hard to define the boundaries of the system are since there is co-authorship across institutions, countries – used co-authorship to draw boundaries.

87 mailing lists, 30k e-mails, 1500 threads.

Social survey: who do you know?

(opt out – each has a name and picture from a public database…. 300 people from co-authors).

How do you know them? When did you meet, how often do you communicate?

10-30 acquaintances – 191 of 373 responded to the survey.

he didn’t look at betweenness (or apparently like Bonaicich or eigenfactor)

Community structure:

Used Newman-Girvan method. (would have been nice if he had drawn a line around them or something – hard to see – as he did on a later slide)

some communities around country of origin, academic affiliation, position (staff,faculty,phd).

co-auth and acq overlap and are one institution and one discipline (i think he said) but are becoming more interdisciplinary and less inter-institutional

cens collab  communities are open fluid, inclusive, and small-worldish (but not based on prestige).

hubs bring different communities together not as interdisciplinary as we have heard

these q occurred to me: the survey – ethical issues with opt out? ethical issues with pictures? did they shift the order? order alphabetical? fatigue? sort in groups

a: they were ordered by institution and discipline, tried other things but this was what worked

In CS they co-author without knowing each other !?!?! that’s cool. In bio they know each other if they co-auth

wrt ethical questions – director of CENS was on the dissertation committee so was all approved.

Comments are off for this post