Loading Events

Faculty Candidate Seminar

IDEs for Ideas

Andrew HeadPostdocUniversity of California, Berkeley
WHERE:
Remote/Virtual
SHARE:

Zoom link

Passcode:  573847

Abstract:  A Jupyter notebook, a preprint from arXiv, and a Stack Overflow code snippet. What do they all have in common?

 

All convey complex ideas from a writer to an audience. However, the tools to create and read them have not kept pace with the complexity of their contents. What would it look like if we had tools as sophisticated as state-of-the-art IDEs (integrated development environments, like Visual Studio or IntelliJ) for authoring and reading these artifacts? What interactions would they provide? What algorithms and models would they need?

 

In this talk, I discuss novel core requirements for reading and writing complex information artifacts, and reify these in the design and implementation of IDEs for Ideas. One IDE lets you click on results in a messy Jupyter notebook and trace the code that computed it. Another brings the meaning of scientific papers to life. The underlying implementations rely on program dependency analysis, version management, and text processing, among other techniques. Come to this talk to learn about the challenges we face using complex information artifacts every day, a vision of how our tools should help, systems embodying this vision with novel interactions and algorithms, and evidence from usability studies validating their design.

Bio: Andrew Head is a computer science postdoctoral scholar at UC Berkeley in the field of human-computer interaction. He holds a Ph.D. in computer science from Berkeley, advised by Björn Hartmann and Marti A. Hearst. His thesis introduced interactive program distillation, the tool-assisted transformation of existing code into sample programs.
For his work on IDEs for Ideas, Andrew has been awarded an NDSEG Fellowship and four paper awards at premier ACM and IEEE conferences. Andrew has researched tool design in collaboration with Microsoft Research, Google, and the Allen Institute for AI. One of his systems, nbgather, is maintained by Microsoft as part of VSCode, and another system will be released soon as a public alpha. Andrew’s IDEs of choice are VSCode for implementing interactive systems, and ‘vim’ with citation management for writing papers. Read about his research at https://cs.berkeley.edu/~andrewhead.

Organizer

Cindy Estell

Faculty Host

Cyrus Omar