Search

Because of the limitations inherent in algorithmic methods of interpreting spectra, library search methods are likely to be used for the immediate future, except for special problems (e.g. Peptide sequence analysis). Nevertheless, the development of such algorithmic methods can be valuable in increasing our knowledge of the decision processes involved when a human analyst interprets a spectrum. For successful application of library search systems, large computerreadable data files are a prerequiste. Such files should preferably contain the complete information from the original spectra. Today computers are being increasingly used for data acquisition in analytical laboratories, hence the production of such library files should soon become feasible as progress is made in defining significant spectral features and assigning balanced weights to them. Expert knowledge and much practical experience in spectroscopy is needed for this task, as well as familiarity with statistical information theory. Systems where computer-aided interpretation of spectral data is used in order to identify the structure of organic compounds will find their main application in fields where samples have to be routinely analyzed without knowledge of their origin, e. g. in the fields of forensic chemistry, analysis of metabolites, and environmental pollution. [ABSTRACT FROM AUTHOR]