2005 IRISH SCIENTIST YEAR BOOK

Home Page

Table of contents

Index by Author

Index by topics

Search


Institute of Technology Carlow
Greg Doyle
Innovation in data mining at the Institute of Technology Carlow

The Data Mining Unit is the newest research group in the Computing and Networking Department at the Institute of Technology Carlow. Data mining research began in earnest in the department in 2002. A new data mining technique, termed Phase Coherent Data Scatter (PCDS), was originally developed in IT Carlow by Dr Norman McMillan several years ago for the specific application of tensiography. Tensiography is the fibre optic collection and subsequent digitisation of light, which has been injected into a pendant liquid drop. The collected light yields a characteristic data set called a tensiotrace. When investigated using PCDS, tensiotraces yield valuable information regarding the physical and chemical composition of the liquid under test.

The data mining research commenced with a single doctoral computing student attempting to experimentally substantiate the hypothesis that PCDS could be adapted to operate as a generic data mining technique. To this end, work was undertaken with researchers in a variety of other academic institutions and with industrial partners. Generic PCDS software, with comprehensive statistical measures and additional functionality, called TraceMiner, was developed. This original work has now been largely completed with successfully concluded and ongoing applications studies on PCDS using TraceMiner in a number of diverse areas. These application areas include:

  • Application of the data scatter technique for the assessment of photometric and wavelength accuracy of UV-visible instruments (in conjunction with Starna/Optiglass). Data gathered from UV-visible equipment can be compared to reference data using the TraceMiner PCDS software and calibration errors can be detected and quantified.

  • Application to micro-Raman spectroscopy for fault detection in silicon wafer fabrication. Silicon wafers are raster scanned with infrared light. This light is collected and the data sets processed using PCDS. Fault detection is possible leading to extensive savings in the fabri-cation process. This work was carried out in conjunction with Trinity College Dublin which works with Intel Ireland Limited.

  • Application to wine quality control. Wine quality control is being investigated using PCDS and standard tensiography with new algorithms inspired by data scatter in development. The overall aim of this ongoing M.Sc. study is to investigate the use of new instrumental quality control techniques and related data mining approaches to providing a quality control tool for wine producers. This study will deliver a software tool that is designed specifically for wine producers and will integrate seamlessly with their existing computer systems. This work is being carried out using TSR Strand 1 funding.

  • Application to telecommunications. This work, among other things, uses PCDS to analyse computer network signals and attempts to identify and quantify errors such as rise time errors and attenuation issues.

Data-diffraction patterns are a new aspect of data scatter that have been identified and are currently under investigation. The diffraction patterns that emerge when data is processed using data scatter tend to be characteristic and identifiable. Analysis and quantification of these patterns is the focus of continued investigation with techniques such as Hough and Radon transforms, among others.

Future work includes further refinements to, and improvement of, the data mining using data scatter and the application of this powerful technique to new disciplines.

Contact: Greg Doyle, Department of Computing and Networking,
Institute of Technology Carlow, Kilkenny Road, Carlow;
Tel: 059 91 70522; E-mail: [email protected]