Uncertainty Modeling of a Group Tourism Recommendation System Based on Pearson Similarity Criteria, Bayesian Network and Self-Organizing Map Clustering Algorithm

Group tourism is one of the most important tasks in tourist recommender systems. These systems, despite of the potential contradictions among the group's tastes, seek to provide joint suggestions to all members of the group, and propose recommendations that would allow the satisfaction of a group of users rather than individual user satisfaction. Another issue that has received less attention is the uncertainty of the memberships ambiguity of a tourist to a group in these systems. This kind of uncertainty is generally caused by the lack of complete information about the opinions of all members in a group and the uncertainty in the process of aggregating users’ viewpoints. The purpose of this research is to develope a group recommendation system through uncertainty modeling. For this purpose, a recommendation algorithm based on Bayesian network, Pearson similarity factor and Self-Organizing Map (SOM) clustering algorithm have been developed. Using the Bayesian network and probabilistic relationships, the existing uncertainties are modeled and their tourism preferences are estimated for each group. Also, according to the relevance parameter in calculating similarity among users, the effect of insufficient information about users in the criteria scoring phase was further reduced, which leads to provide similar recommendations to more similar individuals in a group. Further, tourist attractions and the optimal routes between them are proposed to each group of users via Google map. The results show that in the worst case the value of mean absolute error (MAE) is equal to 1.263, while it is 0.032 in the best case. Also, the success score demonstrates a high level of satisfaction while the maximum and minimum values are 75.353% and 58.509% respectively, which indicates the success of the developed group recommendation system.