Abstract: Aiming at the low accuracy issue of variant mining of design pattern, a variant mining method of design pattern is proposed by introducing clue constraints. , which aims to describe variant clues of design pattern based on the constraint satisfaction problem (CSP). The DPVMC (Design Pattern Variant Mining with Clue) algorithm is given, which introduces clues in two stages: structural feature constraints and timing feature constraints. Taking Proxy, Bridge, Command, and Factory Method pattern variants as examples, a two-stage single design pattern variant mining experiment and an integrated design pattern variant mining experiment are designed. Design pattern mining experiments are carried out through 4 mainstream design pattern mining tools and 4 benchmark systems. The experimental results show that this research method achieves good results.

Key words: design pattern, design pattern variant, clue, constraint satisfaction problem, constraint