There has been increasing growth in deployment of recommender systems across Internet sites, with various models being used. These systems have been particularly valuable for review sites, as they seek to add value to the user experience to gain market share and to create new revenue streams through deals. Hotels are a prime target for this effort, as there is a large number for most destinations and a lot of differentiation between them. In this paper, we present an evaluation of two of the most popular methods for hotel review recommender systems: collaborative filtering and matrix factorization. The accuracy of these systems has become a focus, as more accurate recommendations can lead to increases in profits through various means. Also, given the rapid growth of big data, processing speed to calculate recommendations is an important issue. Using hotel reviews from the TripAdvisor website, we measure the speed and accuracy of these two recommender system methods to determine which method is superior, or the trade-offs between them. The result of the evaluation is a 10.58 times difference in speed of the collaborative filtering method over the matrix factorization recommender system method, but with significantly better accuracy with the matrix factorization method.
Khaleghi, Ryan; Cannon, Kevin; and Srinivas, Raghuram
"A Comparative Evaluation of Recommender Systems for Hotel Reviews,"
SMU Data Science Review: Vol. 1
, Article 1.
Available at: https://scholar.smu.edu/datasciencereview/vol1/iss4/1
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License