join Science and Technology Studies (STS), the Center for Science and Innovation Studies (CSIS), and the Institute for Social Sciences (ISS) for a talk by Premkumar Devanbu and Vladimir Filkov:
When: Tuesday April 14th from 12:10 – 1:30 PM
Where: SS&H 1246 (STS/CSIS Room)
Lunch provided. Please RSVP if you plan to attend.
Premkumar Devanbu, "’Naturalness’ and Empiricism in Software Engineering"
Abstract: Thanks to the emergence & proliferation of open-source software (OSS) projects, software engineering researchers have access to abundant data, on not only <what> kinds of code people write, but also <how> they create, evolve and fix bugs in code. OSS has thus led to a "abundant empiricsm" in software engineering, leading to new, well-grounded insights into software process (how people write code) and software products (what kinds of code actually get written).
I will present some of our findings this area, specially focusing on some new work on the "naturalness" of code, where we have found that code that people actually write is highly repetitive and predictable, thus providing surprising new opportunities to apply statistical NLP techniques such as language models to problems in software development, such as code completion, assistive devices for programmers, code summarization, code retrieval, etc.
Prem Devanbu received his B.Tech from IIT Madras, before your were born, in Chennai, India, and his PhD from Rutgers University in 1994. After working as a developer and a researcher at Bell Labs and its various offshoots, he left Industry to join the CS faculty at UC Davis in late 1997, where he is now Professor of Computer Science. He has published a few papers, graduated a few students, won a few awards, and got a few grants over the years. He even has his own web page.
Vladimir Filkov, "Social Task Networks in Open Source Software Projects"
Abstract: How do distributed groups of
people self-organize to produce artifacts of high quality, like OSS,
Wikipedia, etc.? Are earned trust and spontaneous cooperation essential
determinants of such systems in the absence of rigid organizational
structure? Over the past 3
years, we have adopted a task-oriented, social network perspective to
study organization and productivity in OSS projects. To
do so we were relying on trace data of interleaved social and technical
activities of developers, gathered from software repositories. We
had to develop robust data analytic methods, rooted in network science
and statistical modeling, that could handle heterogeneous data sets
spanning many years in time, and having large amount of variance.
Here, I will talk about
the methods we developed and our results from three separate thrusts of
this work: (1) on measuring contributions among OSS developers, (2) on
collaboration in task networks, and (3) on the temporal congruence of
OSS developer activities and the software call-graph structure.
Vladimir Filkov is an Associate Pro of Computer Science @ UC Davis. His research interests revolve around modeling and data analytics for self-organizing, networked systems, and span the domains of empirical software engineering, bioinformatics of gene networks, and organization of complex social architectures. Prof. Filkov is a member of ACM and IEEE and has been funded by the AFOSR, NSF, USDA, and USFS.