Page 1 of 1

Insights on product labeler problem

PostPosted: Mon Dec 10, 2012 5:27 pm
by samuelchvez
Hi guys, I'm pretty new to AI field, but I'm deeply interested in it.

I'm working in an automated product labeler that should work as follows:

Given a set of indicators (keywords) for a given product, infere what kind of product it is (label it).

I've managed to get almost 8000 string keywords, 500 string labels, and almost 5000 supervised training cases (set(keyword) -> label), and I was willing to use Java Weka / Mulan for this task.

The problem is that my trainer program has been runing for more than 6 hours and it has not yet completed the training ... I need a simplier approach to succeed at this task.

Any insights / ideas?

Have a great day!