Computers are doing more with language now than ever before, and a small army of word crunchers and data twiddlers are hard at work making it all happen.

It’s all on the table, from predicting the stock market to detecting when you’re being a jerk on the Internet. How does it all work?

Linguist Daniel Midgley speaks with computational mastermind Arjun Mukherjee on this episode of Talk the Talk.


Listen to this episode

Or subscribe via iTunes
Click here for more options: How to listen to Talk the Talk


I love going to ACL conferences. That’s the Association for Computational Linguistics. Great people, interesting topics, lots of fun. It’s in Sofia, Bulgaria this year, but sadly I’m not going.

One guy who I’m pretty sure is going is Arjun Mukherjee, from the University of Illinois at Chicago. He’s got three papers accepted, which is quite a feat. I got the chance to ask Arjun about his work, and that’s what this podcast is about. We talk about getting computers to tell a productive discussion from a frustrating flame war automatically. Also, predicting stock market movement. Fascinating!


Show notes

Here are links to the abstracts for Arjun Mukherjee’s papers.

Public Dialogue: Analysis of Tolerance in Online Discussions
http://www.acl2013.org/site/accepted/604.html

Exploiting Topic based Twitter Sentiment for Stock Prediction
http://acl2013.org/site/short/2059.html

TED talk: How algorithms shape our world by Kevin Slavin
http://ed.ted.com/lessons/kevin-slavin-how-algorithms-shape-our-world

A terrific primer on opinion mining (PDF)
http://gate.ac.uk/sale/talks/gate-course-jun12/track-3/module-12-opinion-mining/module-12-opinion-mining.pdf


Show tunes

‘Packt Like Sardines in a Crushd Tin Box’ by Punch Brothers
from the album Antifogmatic

‘Money (That’s What I Want)’ by Barrett Strong
from the album Money (Single)