"Data mining in the cloud" : Ex-Microsofties Launch $500 'Meaning Machine' For Large Data Sets

Dominic Pouzin is a worldly, smart guy. After doing school and internships in France, the UK, South Korea and India, he moved to Atlanta where he took a job in the NBC (Nuclear, Bacteriological, Chemical) protection field. "I designed statistical models, programmed robots, and implemented access control stations for nuclear plants," Pouzin says on LinkedIn.

He left that field to become a software engineer at Microsoft and this week launches his own startup company with a collection of former Microsoft co-workers. The new company is called Data Applied and offers very affordable "data mining in the cloud" - it applies automated algorithms to large sets of data in order to extract patterns, preconditions and outliers.

Commercial licenses cost under $500 and are aimed to bring the kind of power to sales, marketing, engineering, social sciences or non-profit organizations that only banks and insurance companies used to be able to afford.

There's a Data-Centric World Emerging

Data visualization tools are not uncommon online and there's a growing number of large data sets available for analysis these days. Microsoft unveiled a marketplace for data sets last month called Dallas. GigaOm's Stacey Higginbotham says data represents a big part of Microsoft's hope for the future. "Microsoft isn't just selling software anymore," she writes, "it's selling itself as a clearinghouse for information." Higginbotham points out similarities between Dallas and a startup we've covered here extensively called InfoChimps. Amazon started offering large sets of public data through its Web Services division last year as well. We've got high hopes that social networking data will be made more available for aggregate analysis, as a bird's-eye-view of society's interactions.

We've written here before about how all of this data to be analyzed may or may not be put to good use (Web 3.0 Might Be Really Stupid) and Nicholas Carr recently wrote a provocative blog post about the consequences of data-driven recommendation on free will and human thought.

You've Got the Data, Now What?

Countless organizations have data sets of their own but that doesn't make extracting valuable information from them easy. Microsoft itself unveiled a similar looking service last month called Pivot, but it's invite-only for now and not thoroughly explained.

We're familiar with DabbleDB, but Pouzin says Data Applied takes things much further than data visualization. "Simple reporting and dashboards are boring," Pouzin says. "They force analysts to manually visualize all possible combinations in the hope of finding some interesting facts. We do that automatically!"

Data Applied performs a whole lot of data visualization functions as well, it's social and offer a multitude of related features while running on one or more computers.

Can it really pull unforeseen patterns and meaning out of large sets of data, though? That sounds like a tall order to fill, but if enough numbers can be crunched that the mind of a human consultant is in fact unneeded - then Data Applied could be bringing to market a very valuable service.

Copyright 2009 ReadWriteWeb. All Rights Reserved.

Source : NYT, 1/12/09