Integral Solutions Ltd, Basingstoke, Hamsphire-based software designer, has come up with a new data mining tool called Clementine. It will enable all of you out there dying to make use of neural network and rule induction programming to use this technology without the hassle of having to learn the programs yourself. According to Integral Solutions, Clementine is extremely user friendly because it uses visual programming. Users are able to select graphical icons representing data sources and operations and then position, connect and edit them accordingly. Clementine will then read, process and display the data. The software will enable the user to make forecasts, predictions, estimations and decision support, the company says. Once data has been extracted and tested it can be exported as source code modules for progamming languages. Clementine has been on beta test since Easter. The BBC is at present using it to estimate audience viewing figures and predict the effects of changes to its scheduling. The neural network engine will estimate the popularity of the television programmes and the Rule Induction will classify the data into share ranges and illustrate the effects of factors on programme popularity. In this way under- and overperforming programmes can be identified as well as the factors that determine what makes a programme popular to within 96% accuracy. Clementine connects to Oracle, Ingres, Sybase and Infobase relational databases and can import file formats to cover other systems. It runs under Unix and Digital Equipment Corp’s VMS. Colin Shearer, a director of Integral Solutions, says that the company is currently working on a new version based on parallel processing that will be ready by the end of summer. Version 1.0 costs #12,500 for the basic configuration with a single user licence.
