A Tourism Recommender System using CF approach and Matrix Factorization
Authors/Creators
- 1. Safashahr Azad University, Fars, Iran
- 2. Head of Software Engineering Department, Safashahr Azad University, Fars, Iran
Description
Abstract
With the development of the Internet, technology, and communication devices, the production of
tourist data at all levels (hotels, restaurants, transportation, heritage, tourism events, activities, etc.)
has multiplied. However, the list of possibilities provided by these web search engines (or even
specialized tourism sites) and offered to tourists, can be abundant, which causes information
overload. Many recommender systems have been created to help tourists plan their trips and help
them find the information they are looking for. In this paper, first, the different approaches of
recommender systems in the field of tourism are reviewed. Then, the proposed method of this
study is presented. Primarily, similar users are identified based on their preferences and priorities
and using cosine distance. If the same users have not rated the visited places, this vacancy (blank
space) of rating will be filled using the ALS factoring matrix. Then, to recommend the next
location to the target user, the similar users to the target user is calculated using the cosine distance,
and the weighted average for the 3 top locations is calculated, and the location with the most weight
is recommended to the target user. According to the results presented in Section 4, the proposed
method has more accuracy, recall, and F1 in the recommendation than the CF method.