In order to improve the efficiency of retrieving models from enterprise model library

a similarity algorithm based on weighted flow relationship is proposed

regarded as WF. Firstly

this paper designs a weighted flow relationship set generation algorithm. The generation algorithm traverses all nodes and flow relationships except the output place by breadth-first search method. The weight is assigned to each flow relationship to generate a weighted flow relationship set. Secondly

the WF algorithm calculates the weighted transition set based on the weighted flow relationship set

and calculates the similarity according to the WF similarity definition. Finally

through experimental and analyzing the time complexity of the algorithm results show that the proposed algorithm is correct and unique to determine the result and the algorithm has good performance.