Rubberband Wound

http://en.wikipedia.org/wiki/Natural_language_processing

Computer language

specific
well-defined

Natural language

ambiguous
context sensitive

John saw the man in the park with the telescope.

Who has the telescope?

Processing natural language

stemming
breaking apart sentences
tokenizer

Named Entity Detection

recognizing chunks in text

person
date
money

Mr Jones went shopping on 31st May. He spent $3.15.

You will probably never be able to expect 100% accuracy in NLP.
Problem areas include companies named after people

Robert Half, Inc.
Thompson West

By Garrick Van Buren
 I hit the back button and lost all my notes. That’ll teach me to try to do this in the browser.

Smart Defaults - when a user asks for something, see if it makes sense for 80% of users. Most of the people who speak up are edge cases, in the other 20%.
What […]

By Ben Edwards and Ross Niemi - 1100-1150
Slides are online.

Agile can be a hard sell, and it’s our fault because we’ve always wanted to get all the specs upfront from clients.
Benefits are Clear

Agile methods reduce project costs
Agile methods get projects to market faster
Agile methods create higher quality software

Communication, Responsiveness, Transparency, Growth

Keeps people on the same […]

by Robert Fischer - 1000-1050

The Pragmatic Programmer - if you are repeating tasks, automate them.
How can we translate customer specs and requirements to documentation?
FIT - Framework for integrated testing
Story Test - Based on Unit Test idea

Customer produces expectations and communicates them to a developer
Hopefully it is documented for future reference - if it’s not documented […]

I’m at MinneBar right now, the turnout is huge. It’s been said that this is the largest group at a BarCamp other than one in India.
 I’ll be posting notes from panels during the day.