Abstract: In order to solve statistical skew and efficiency problems, it presented a method based on the ER(Entity-Rela- tionship) concept model. hhe method takes relationship sets of an ER model as the core, and applies expanded SQI_ sta- tistical primitives of relational databases to produce multi-relational frequent patterns based on data and interestingness constraints which are provided by users, such that the patterns' number may be greatly reduced. Not only no physical joining of relational tables is needed, but also no statistical skew is produced. hhe comparison with some related research works explains the effectiveness and correctness of the method, which is performed by making use of the relational data- base management system and the ER model.

Key words: Data mining, Multi-relation, ER model, Frequent pattern