COMPUTER-STORAGE AND RETRIEVAL OF GENERIC CHEMICAL STRUCTURES IN PATENTS .17. EVALUATION OF THE REFINED SEARCH

Results are presented for the atom-level search, called the refined se arch, for matching components of generic chemical structures. The refi ned search is the last and most discriminating search strategy used by the Sheffield generic structures system and is performed after the fa ster screening stages, bit screening, and reduced graph screening. It operates on the real atom representations of components of the generic structure which are defined by reduced graph nodes, nodes which repre sent aggregates of atoms of the original chemical graph which are stru cturally similar. The nature of generic structures means that parts of a structure may be expressed in terms of real atoms and bonds and par ts in terms of homologous series identifiers such as alkyl. The fundam ental problem concerning the refined search is determining equivalence s between these two types of representation. Search results are presen ted for four query database searches against one file database. The me thodology and results of the screening searches are also briefly descr ibed.