A Recommendation Algorithm that combines Rating Data and Review Text

Inhalt:
There exit a huge amount of rating data and user comment in the E-commonce systems. However, it is still a difficult problem for the recommendation system to make better use of them altogether. In the literature, some use only ratings, and some use only reviews. Recently some recommendation models try to learn from comment text and rating matrix both. However, they only focus on the topic of the comment text, and do not make full use of the comment.