In the recommender system, the most important is the decision-making solution to consulte for user. Depending on the type and size of data stored, decision-making will always be improved to produce the best possible result. The main task in implementing the model is to use methods to find the most valuable product or service for the user. In this paper, we propose a new approach to building a multi-user based collaborative filtering model using the interaction multi-criteria decision with ordered weighted averaging operator. This model demonstrates the synergy and interplay between user criteria for decision making. The model was evaluated through experimentation with the multirecsys tool on three datasets: MovieLense 100K, MSWeb and Jester5k. The experiment illustrated the model comparison with some other interactive multi-criteria counseling methods that have been reserched on both sparse datasets and thick datasets. In addition, the model is compared and evaluated with item-base collaborative filtering model using the interaction multi-criteria decision with ordered weighted averaging operator on both types of datasets. Consultancy results of the proposed model are quite effective compared to some traditional consulting models and some models with other operator. This counseling model can be applied well in a variety of contexts, especially in the case of sparse data that will result in improved counseling. In addition, with the above method, the user-base model is always more efficient than item-base on all datasets.