Extraction of information from the text of chemical patents. 1. Identification of specific chemical names

Much attention has been paid to translating isolated chemical names into forms such as connection tables, but less effort has been expended in identifying substance names in running text to make them available for processing. The requirement for automatic name identification becomes a more urgent priority today, not the least in light of the inherent importance of patents and the increasing complexity of newly synthesized substances and, with these, the need for error-free processing of information from patent and other documents. The elaboration of a methodology for isolating substance names in the text of English-language patents is described here, using, in part, the SGML (Standard Generalized Markup Language) of the patent text as an aid to this process. Evaluation of the procedures, which are still at an early stage of development, demonstrates that even simple methods can achieve very high degrees of success.