Abstract
Although recommender systems (RSs) strive to provide recommendations based on individuals' histories and preferences, most recommendations made by these systems do not utilize location and time-based information. This paper presents a travel recommender system by integrating the Imperialist Competitive Algorithm (ICA) and Fuzzy C-Means (FCM) Clustering algorithm. Compared to similar studies, this recommender system takes into account more POIs, including location, number of visits, weather conditions, time of day, user mood, traffic volume, season, and temperature. The effectiveness and accuracy of the proposed method are assessed using the Flickr dataset, indicating that it is able to provide effective and accurate recommendations that are compatible with the user's interests and the current status of his/her visit. Results showed that, precision and Mean Absolute Precision (MAP) in the proposed method have been grown 23.6% and 23.72% in comparison to Popularity Rank, 28.98% and 19.67% in comparison to Classic Rank and 18.66% and 19.67% in comparison to Frequent Rank methods. Also, Mean Absolute Error (MAE) index in proposed method has been improved 60.71%, 64.51% and 56% in comparisons to the Popularity Rank, Classic Rank and Frequent Rank methods respectively.