Artificial intelligence methods for performing a query in search engines require effective solution of the task of automatic processing of texts in the database. One of the important tasks of implementing machine learning is the development of a rule system. This article discusses the problem of coreference calculated from preliminary studies aimed at automatic determination of meaningful distances expressed in Uzbek language texts. Basic rules and stages of solving the coreference problem are also analyzed. Explanations of a number of terms such as anaphoric relations and their main types, the main used anaphora, antecedent, cataphora, clustering, singleton, etc., which create the coreference problem, are given.