A New Similarity Method for Tourism Recommendation Systems

Abstract

In this paper, we proposed a new similarity method to use in tourism recommendation systems. Recommendation systems highly depend on the existence of a similarity measure used to identify similar items. In tourism products such as hotels, trips, packages are all hard to judge for their similarity. The proposed method is simply based on user defined weights to calculate similarity. First, we represented each product as a vector and then weighted by user defined scores. Then it uses cosine similarity to measure similarity between items. We evaluated our method using a dataset created by the travel expert. Our experimental results indicate that the proposed method achieves a significant improvement in terms of mean average precision (MAP). We conclude that the proposed method is a promising approach for improving the performance of tourism recommendation systems.