Annotation: To formulate optimization tasks in the Electronic Policy Document Management System, an enterprise can be presented as a set of departments performing certain functions. At large enterprises there is a constant exchange between the individual departments of the enterprise, between the document databases of departments. The paper examines solutions to several document database- related problems. Here, first of all, we consider the problem of choosing a model capable of setting the relationship between the query execution time and the amounts of documents, grouping of policy documents according to their sources of creation and evaluation of the advantages that this grouping can offer. An experimental table showing the relationship between data size and query execution time is given to solve the problems set. The least squares method is used to select linear or hyperbolic models used in this table. The closeness of the model of this relationship to a linear dependence is used in evaluating the advantages of partitioning the database into natural groups. It is known that policy document databases store documents of local or international importance that are created and recommended for implementation by local or international organizations. To improve the work, the documents in the database can be grouped by the degree of their importance, by the date of entry into the system, by the organizations that created the document. The paper presents a method of grouping policy documents by the source of their creation. Some numerical characteristics indicating the advantages of this grouping are given

Keywords: policy document database, data source, query time, amount of information, additional delays, mean-square difference, linear model, hyperbolic model, natural division