Dan Zambonini needed a way to automatically classify incoming tweets regarding customer service simply as positive or negative. He shares how he used naïve Bayesian classification and a couple of PHP classes, removed ‘noise words’ and created an interface that pulls tweets from Twitter. Further tweet tweaks improved the results, and he ended up with very good accuracy, and an automated analysis tool whose results should improve even further over time. Dan’s clear writing style is easy to follow and I found it fascinating and relevant, as the mining of the data tsunami that is Twitter is a 21st-century gold rush.
Digital Gold Rush: Mining Twitter
Previous post: Geek Travel In China
Next post: HTML5 Video Demo
