Embodiments of the disclosure provide a method, apparatus, and system for identifying data tables. The method comprises acquiring a first dependency relationship between a plurality of data tables; collecting statistics on a path length and a path number of the data tables based on the first dependency relationship; acquiring a second dependency relationship between one or more fields in the data tables; determining importance coefficients of the one or more fields based on the second dependency relationship; determining a degree of association between the data tables by using the path length, the path number, and the importance coefficients; and identifying the data tables based on the degree of association.