"Just Enough NLP with Python" is a little seminar I gave to the Parse.ly team about doing NLP with Python/NLTK and some of the problems with entity idenitfication in particular.
Slides can be viewed in compiled form at:
http://pixelmonkey.org/pub/nlp-training/
Note that the slides can be controlled as follows:
- Advance forward / back with the forward and back keys, or left click / right click of the mouse
- Press
cto get the "controls", which also allows you to skip slides and switch to outline mode - Outline mode includes some notes not included in the slidedeck, and also allows you to easily copy/paste examples into your own interpreter
I suggest you run through the slides in slide mode, and then review them in outline mode, doing examples from your own interpreter. That's how I tended to do things when I physically gave the presentation. Of course, you can also contact me on Twitter at @amontalenti if you want to see if I might be giving the talk nearby you sometime soon :-)
Using Python, of course. It's turtles all the way down.
I wrote the slides using reST, and specifically Docutils support for S5 export. Scripts are included to compile the presentation from the index.rst file and also to allow development of new slides with live recompilation using pyinotify (Linux systems only). See build.sh and monitor.sh for more information.