Abstract: For strengthening accuracy of user model construction and matching in adaptive cartographic visualization system, this paper has put forward quaternary interaction user model and similarity matching algorithm based on probability. Quaternary interaction user model, which adequately considers “human” “machine” interaction character in ACViS, can more roundly reflect characteristic for user using ACViS. Based on similarity calculation method in common use, similarity matching algorithm based on probability has been put forward. This paper has introduced the basic idea and process of algorithm in detail, and has given an example to validate. It is tested that this model matching algorithm is easy to be realized and can match more effectively, which considers actual circumstance of ACViS, and improves the accuracy of model matching, so, it provided a new approach for user model matching.