In this article, a new model is proposed for tourism recommender systems. This model recommends tours to tourists using data-mining techniques such as clustering and association rules. According to the proposed model, tourists are initially clustered. Self Organize Map (SOM) algorithm is used for determining the number of clusters and the clusters are created by K-means algorithm. Then, the clusters are analyzed and validated considering Quantization error, Topographic error and Davies-Bouldin error parameters. This model is implemented using two methods; according to the first method, recommendation is made based on tourists' location, and in the second method this is done based on tourists' behavioural patterns in the past. The results from evaluating the model using PearsonCorrelation show that recommendations based on the behavioural patterns are closer to tourists' interests.

In major e-commerce recommendation systems, the number of users and items is very