A general method for matching arbitrary logical statements in mechanized retrieval systems

A general method for matching specifications in mechanized retrieval systems is presented. It is based on the principle that, with the initiation of the matching operation, to each characteristic or descriptor a unique code consisting of a power of two in binary notation is assigned. These codes may, if so desired, be interpreted as implicit weight coefficients of the descriptors. They are used in matching operations between sets of characteristics. The matching process produces during its progress criteria as to the relations between involved sets, i.e., a numerical critèrion of relevance of the answer to the query, which may be utilized to control the course of the operation. In a second step, new secondary codes are derived mechanically from those assigned initially. The function of the secondary codes is to transform the original statement into the second canonical form, consisting of the disjunction of mutually exclusive basic conjunctions or propositions. These secondary codes are employed to generate by logical multiplication, addition, and negation unique code representations for arbitrary compound statements or propositions, which representations in their turn are employed for matching the statements. The process generates criteria in analogy to those obtained in the first step of matching sets of characteristics which now permit the control of the matching operations between arbitrary propositions. It should be mentioned that no significance is assigned a priori to the codes. Their relative significance is restricted entirely to the matching operation actually in progress.