Abstract: A computer-based method and apparatus for identifying hazardous materials based on observable physical properties and signs and systems of exposure to the hazardous materials. The observable physical properties are obtained by preprocessing plain language data to obtain normalized descriptors and modifiers pertaining to a plurality of properties and signs and symptoms. The normalized descriptors and signs and symptoms are presented to a user in lists for comparison with a sample in the field. Tightly integrated with identifying the hazardous materials are obtaining response data, obtaining controlled materials of which the hazardous materials may be precursors, and predicting reactions of pairs of hazardous materials. The preferred embodiments of the invention are particularly useful to responders to a hazardous waste incident.