Review-Based Service Profiling and Recommendation
Toshihiko Yamasaki, Masafumi Yamamoto, Kiyoharu Aizawa · 2016
Finding desired services (e.g., restaurants) from a number of candidates is a difficult task. We usually look at the scores given by other customers or read their reviews for decision making, but it is time consuming. This paper presents review-based service recommendation using both a bag-of-words model and a skip-gram-based model. Namely, instead of checking the scores or review comments, a review is given to the system as a query, and the services to which similar reviews are given are recommended. We evaluated our algorithms by using Yelp's Academic Dataset. The experimental results show that the percentage of finding the corresponding restaurants as the top-1 candidate was 25% and that in the top-10 ranking was 51%.