Searching for regularities in a Raman spectral database

The paper describes application of artificial neural networks and statistical methods to analyse a Raman spectra database and to use the results of the analysis in structure elucidation process. The statistical approach enables automatic generation of rule knowledge bases, which can be used to interpret Raman spectra of an unknown compound and, finally, to recognise substructures present (or absent) in the compound's molecule. The artificial neural network approach generates networks which are able to analyse Raman spectra, thus recognising the presence (or absence) of certain substructures in the compound's molecule. (c) 2004 Published by Elsevier B.V.