Abstract
A technique of a database constructing which contains materials of the Russian Empire censuses, by the example of census returns of XVIII-XIX centuries is considered. The database is based on the digitization of archival materials census made in Excel-format. An algorithm for recognition of semantic structures in the records of census documents, the detection of attributes and rules for filling the relational tables are described. The algorithm is given for the specific census (4 and 5 revisions), but these algorithms can be also developed for other revisions and it will make possible to build a continuous chain of historical census lists, translated into a single database format. Such a chain can provide meaningful assistance in genealogical research.