1987 Volume 36 Issue 5 Pages 339-341
A pattern recognition method for elucidating functional groups of compounds from their IR spectra has been investigated by utilizing spectral data compiled in the Spectral Data Base System developed in our laboratory. Spectral data of about 9500 compounds were used, and error ratios in the elucidation of six functional groups (C=O, C=C, C≡N, C-O-C and NO2 groups, and benzene ring) by using a learning machine method were evaluated. As a result, it was found that all functional groups except C=C could be elucidated with high probabilities above 85%. Thus the present method has a high performance when applied to the elucidation of functional groups from IR spectra.