Abstract

The possibilities and limitations of chemical reaction names as the basis for retrieving literature dealing with reaction types are briefly reviewed, as well as the capability of a mere substructure search in a file of reactants and products. It is demonstrated that reaction indexing ideally should be based on the identification of the reaction sites of the molecules involved, as well as on the representation of various kinds of relations between these reactive centers. The GREMAS reaction indexing of IDC is discussed as an example. Before a chemist can request information on reaction types that appear promising for a certain target molecule, he must be aware of their existence. The task of computer-assisted synthesis design is to submit suggestions of conceivable synthetic pathways to a target structure and so to alert the chemist to syntheses of which he had not yet been aware. The “reverse retrieval” technique is demonstrated to constitute a fundamental principle for this kind of information supply. If sufficiently developed and assisted by an effective system for the documentation and classification of reaction types, it can provide suggestions for syntheses at a considerable efficiency without requiring completion by additional principles. The IDC system of (Sub-) Structure search seems to be able to serve this purpose too, due to its flexibility, economics, and precision. The reverse retrieval offers solutions to a number of problems hitherto unresolved in information science.

Article Outline

• References