The Evolution of an Automated IR Spectra Interpretation System

Commercial spectral identification and search systems are intended to be used by persons who are not expert spectroscopists. Most pure search systems focus directly on the best matches to spectra that are contained in the library. However, an expert narrows the scope of a search by first identifying patterns in the spectrum assigned to its functional groups or structural units. Then, when searching the library for matching spectra, attention is focused on compounds containing these structural units. The effect of the structural unit prefilter on the library search is to eliminate materials that coincidently match the peaks of the unknown but are chemically unrelated. A significant factor in this search mode is the guidance the searcher receives when the spectrum is not contained in the search library. The evolution of a spectrum recognition system that follows these procedures is described as one of the earliest examples of a commercially successful expert system.