The article proposes a method and algorithm for constructing an informative phase, which is one of the most important issues of symbol recognition. All the concepts related to the problems of symbol recognition have been attempted to be expressed in a logical perfection, mathematical clarity and simplicity. The main issue of classification of symbols⁇, the formation of the phase of informative signs, the determination of the levels of importance of the objects and symbols of the study sample were thoroughly analyzed. The Fisher functional elements used to determine the values of the proximity function in the informative signs phase Upper and lower bounds of the similarity function between objects are identified and supporting data is provided.