Fundamentals of Structure Elucidator System

Abstract

This chapter starts with a concise description of the expert system Structure Elucidator flow diagram (selection of possible substructures → structure generation → selection of the most probable structure based on NMR spectrum prediction) and explains how the system works. The fundamental concepts making up the logical basis of spectroscopic structure elucidation are then described. Special attention is placed on an explanation of the axiomatic nature of the initial information used to logically infer the structure of an unknown. The following three groups of axioms (statements) are distinguished: (i) axioms of characteristic spectral features in 1D NMR and IR, (ii) axioms of 2D NMR (HSQC, HMBC, COSY, etc.), and (iii) axioms used for the assembly of a structure. The properties of the information used for structure elucidation from 2D NMR data are summarized. It is concluded that to solve real-world problems, Structure Elucidator must be capable of deducing correct structures from a fuzzy, incomplete, and frequently contradictory set of axioms composing the initial information. It is noted that an axiomatic approach is a cognitive basis not only for CASE, but also for organic qualitative analysis. The knowledge base of Structure Elucidator consisting of factual and axiomatic knowledge is described and three methods of NMR spectrum prediction included in the system (incremental, neural nets, and HOSE code based) are discussed.