IBM points predictive analytics at social Web
They're even tracking those emoticons
The next time you choose an emoticon to spice up a comment about a company, product or service online, know this: There's a chance that a software program is capturing the sentiment expressed by your smiley or frowny for analysis.
The capability is part of a new release of IBM's SPSS Modeler text mining and analytics software, which also allows customers to parse information from social media sites and feeds.
Emoticons' meaning tend to be obvious, but written comments aren't always so, due to industry-specific jargon and slang. To that end, the new release includes 180 "semantic networks" with taxonomies for verticals like life sciences or consumer tech. For example, in a banking context, the software would recognize that a "variable rate mortgage" and an "adjustable rate mortgage" are the same thing, IBM said.
Many companies already use demographic information, sales data and other structured sources to build predictive models in an effort to understand customer behavior and connect with new ones. IBM contends that integrating information from the social Web will add further insight.
SPSS customer Money Mailer, a direct marketing services provider, definitely sees the potential of social media in its business, but is still in the early stages of figuring out how to use it, said John Gramata, vice president of marketing.
Historically, Money Mailer has used SPSS tools for purposes like creating "look-like" customer models, he said. He cited a regional garden center that had a dedicated customer base but was looking to add more. Money Mailer took a sample of information from the company's "loyalty club" database and developed a model for spotting likely targets aggregated from other contact databases.
Money Mailer has an in-house statistician that works with SPSS, but the software's interface has proven easy for some business analysts to work with, Gramata said.
While generally happy with the software overall, Gramata is looking forward to expected speed improvements in the new release, as very large data sets tended to result in a performance hit. "It's always been a fast tool but ... any kind of improvements in speed is definitely a welcome enhancement."