Data mining became a passion for Themos Kalafatis a decade ago when he was working on his masters degree at the University of East Anglia in England, far from the madding dot com bubble. He moved on to jobs in predictive analytics at HP and Intracom and adopted social media in 2007. Currently an independent consultant, Themos blogs on predictive analytics over at our sister site Smart Data Collective and at his own blog Life Analytics.
But it was his post here on Twitter keywords for marketing predictives that drew 6200 views and 16 comments that makes Themos this week's Blogger of The Week.
"The format of tweets is ideal for text mining," he says. "The 140 character format is great for analyzing group tweets, people and thought clusters". Using cluster analysis he identified similar groups on Twitter based on their biographies and extracted words that each group is associated with. A big plus for those who seek to monetize Twitter and strengthen its marketing metrics.
"One can find associations between brands and specific keywords, find out what people want and don't want and extract what they value most," he adds. "And then find similar groups of online behavior and transfer that to online marketing and a strong web presence."
The possibilities for expanding his methodology seem endless, much as Einstein said there are no points in time. Congratulations, Themos! Thanks for bringing fresh ideas and comment generating themes to SMT's online communities.