KNOWLEDGE-BASED INTERPRETATION OF GAS-CHROMATOGRAPHIC DATA

The interpretation of analytical measurement data, including data from gas chromatography, is based on matching usually newly acquired data with reference data that were recorded earlier with known substances. This procedure can be automated by computer systems using purely numerical algorithms. In general, this procedure is not sufficient to obtain reliable results. The analyst must use chemical and other knowledge to verify substance identifications gained from reference matches. Often several different measurements of one sample have to be carried out and interpreted to meet the requirements of reliability. In order to support verification, the introduced knowledge-based system WANDA can process numerical data as well as symbolically represented information. Since uncertain knowledge plays an important role in the 'manual' interpretation process, the system has been designed to handle also uncertain facts and rules. The WANDA expert system prototype provides assistance in analyzing traces of organic substances, preferably in water samples. Users obtain validated substance identifications by combining knowledge from different sources and an integrated interpretation of related chromatograms. The system has already been used in practice for testing purposes.