Increasing The Predication Accuracy Of Recommendation for The New User Cold-Start In A Recommender System By Using Shuffled Frog Leaping Algorithm

Recommender systems help users to choose the right ones based on their interests. The Cold-Start issue is one of the problems of recommender systems that have a great impact on their performance. Cold-Start problem includes three types which are: new community, new items and new user. When a new user enters or present in the system and does not reach high records, so the system does not have enough information about his interests and tastes to make recommendations, then the new user Cold-Start problem occurs. In this research, by placing users with the most similarity in the same groups and proper clustering, we have tried to make the suitable recommendation for the Cold-Start User based on the taste of the neighboring users. In this method, using C-Means fuzzy clustering and fixing its problems, such as determining the number of clusters to be identified in advance, aswell as falling at the local optimal points, is trying to improve the prediction accuracy. The XMeans algorithm was used to determine the optimal number of clusters, so the shuffled frog-leaping algorithm was used to avoid falling at local optimal points. The proposed method implemented on the MovieLens dataset, which the results showed an increase in the accuracy of prediction compared with the K-Means, C-Means, and K-Medoids algorithms.