When working with independent databases, the problem of partial automation arises, which consists in the impossibility of obtaining data from independent databases with a previously unknown scheme. The problem consists of two aspects: the lack of semantic connection between independent databases, and the dynamics of the semantics of schemes. The goal is the development and study of algorithms for constructing integrated queries for data sampling to heterogeneous distributed database management systems. To test the functionality of the algorithms, software and database of databases, which provides the ability to access independent databases, as if they were a single DBMS, but without full replication of the merging databases.