Abstract
Together with the prevalence of e-commerce and online shopping, recommender systems have been playing an increasingly important role in people’s daily lives in terms of discovering their potential preferences. Therein, users’ preferences are mostly reflected by their online behaviors, specially their evaluation towards particular items, e.g., numeric ratings and textual reviews. Many existing recommender systems focus on using item ratings to determine users’ preferences, while others provide approaches using textual reviews instead. In this work, via a case study on the Amazon movies data, we compare the recommendation results when using ratings or reviews, as well as that of combining both.