I love the Siri Service on Apple's new 4S phones! I want to know what makes it tick, and how it decides how to respond.
- gather Siri questions and responses from actual users
- publish an open dataset of questions and responses for use in making future Natural Langauge Procressing (NLP) applications, based on what questions people are asking in the Real World
- create a parsimonious, mostly-compatible open interface that answers questions like Siri would and does.
- Find the hidden easter eggs in Siri, and suggest places where it can do a better job meeting user goals.
- Speech recognition. For purposes of this project, we assume the questions are in English text (already transcribed)
- http://ask.opensiri.com and describe your experiences with Siri.
- verify questions and responses at http://ask.opensiri.com/verify
- code checkouts at https://github.com/gregglind/OpenSiri/
- SiriIRL: analyze the data (http://ask.opensiri.com/dump)
- OpenSiri: a compatible interface that acts like Siri
- OpenSiriWeb: GAE/AppEngine website
- users, users, users! Know people who use Siri? Have them report!
- designs for website, forms, architecture
- javascript front-end designers.
- Google AppEngine (gae) help.
- lead scientist / statistician for doing heavy lifting on models
- iphone client, to make capture of responses and reporting easier
- help refining the research questions
- bug-testing and bug squashing (https://github.com/gregglind/OpenSiri/issues)
Diverisity
- OpenSiri wants participation from coders and non-coders of all experience levels.
- "We welcome people of any gender identity or expression, race, ethnicity, size, nationality, sexual orientation, ability level, religion, culture, subculture, and political opinion." (http://www.dreamwidth.org/legal/diversity)
- no dev in master, please
- file an issue report, if appropriate
- use git topic branches, with your issue
- make a pull request
- html5boilerplate
- underscore, jquery, and friends
- appengine / python
Gregg Lind <gregg.lind at gmail.com>