Opinion Recommendation Using A Neural Model

Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 1626–1637 Copenhagen, Denmark, September 7–11, 2017. c⃝2017 Association for Computational Linguistics Opinion Recommendation Using A Neural Model∗ Zhongqing Wang†,‡ and Yue Zhang‡ †Soochow University, Suzhou, China ‡Singapore University of Technology and Design, Singapore wangzq.antony@gmail.com, yue zhang@sutd.edu.sg, Abstract We present opinion recommendation, a novel task of jointly generating a review with a rating score that a certain user would give to a certain product which is unreviewed by the user, given existing re- views to the product by other users, and the reviews that the user has given to oth- er products. A characteristic of opinion recommendation is the reliance of multi- ple data sources for multi-task joint learn- ing. We use a single neural network to model users and products, generating cus- tomised product representations using a deep memory network, from which cus- tomised ratings and reviews are construct- ed jointly. Results show that our opinion recommendation system gives ratings that are closer to real user ratings on Yelp.com data compared with Yelp’s own ratings. our methods give better results compared to several pipelines baselines. 1 Introduction Offering a channel for customers to share opin- ions and give scores to products and services, re- view websites have become a highly influential in- formation source that customers refer to for mak- ing purchase decisions. Popular examples include IMDB.com on the movie domain, Epinions.com on the product domain, and Yelp.com on the ser- vice domain. Figure 1 shows a screenshot of a restaurant review page on Yelp.com, which offers two main types of information. First, an overal- l rating score is given under the restaurant name; second, detailed user reviews are listed below the rating. ∗This work has been done when the first author worked at SUTD. Figure 1: A restaurant review on Yelp.com. Though offering useful overview and details about a product or service, such information has several limitations for a user who has not used the product or service. First, the overall rating is gen- eral and not necessarily agreeable to the taste of an individual customer. Being a simple reflection of all customer scores, it serves an average cus- tomer well, but can be rather inaccurate for indi- viduals. For example, the authors themselves of- ten find highly rated movies being tedious. Sec- ond, there can be hundreds of reviews for a prod- uct or service, which makes it infeasible for ex- haustive reading. It would be useful to have a brief summary of all reviews, which ideally should be customized to the reader. To address the limitations above, we propose a new task called opinion recommendation, which is to generate a customized review score of the product that the user is likely to give, as well as a customized review that the user would have written for the target product, if the user had reviewed the product. The proposed opinion recommendation task is closely related to several existing lines of work in NLP. The first …

Content truncated to 500 words