Thinking Machines Corp, the company that staked its future on mining very large databases after being forced to ditch its parallel supercompter hardware a couple of years ago (CI No 2,809), has brought its Darwin data mining to the web with the latest releases of the software. Darwin was originally designed for the company’s parallel Connection Machines, which it decided to ditch after a protracted period in Chapter 11 bankruptcy, to concentrate on data mining. It now runs on IBM Corp, Digital Equipment Corp and Sun Microsystems Inc Unix machines. Darwin 2.1 adds the ability to export predictive models in Java, to enable data mining over the web. Models can also be exported in C and C++, which the company says will enable users to integrate data mining results into their existing custom applications, or embed them into SQL queries. Other new features in the release include faster and more robust data mining algorithms, and the ability to export results into Microsoft Corp Excel. The company says it has responded to considerable demand from customers to be able to use the Darwin predictive models over the web. Darwin 2.1 starts at $30,000 for an annual usage fee, and is available immediately on Sun Solaris. A Hewlett-Packard Co and IBM version will be out by the end of the year.
