Attribute-aware Sequence Network for Review Summarization

Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, pages 3000–3010, Hong Kong, China, November 3–7, 2019. c⃝2019 Association for Computational Linguistics 3000 Attribute-aware Sequence Network for Review Summarization Junjie Li1, Xuepeng Wang1, Dawei Yin1, Chengqing Zong2,3,4 1 Data Science Lab, JD.com, Beijing, China 2 National Laboratory of Pattern Recognition, Institute of Automation, CAS, Beijing, China 3 University of Chinese Academy of Sciences, Beijing, China 4 CAS Center for Excellence in Brain Science and Intelligence Technology, Beijing, China {lijunjie46,wangxuepeng1}@jd.com,yindawei@acm.org,cqzong@nlpr.ia.ac.cn Abstract Review summarization aims to generate a condensed summary for a review or mul- tiple reviews. Existing review summariza- tion systems mainly generate summary only based on review content and neglect the au- thors’ attributes (e.g., gender, age, and oc- cupation). In fact, when summarizing a re- view, users with different attributes usually pay attention to specific aspects and have their own word-using habits or writing styles. Therefore, we propose an Attribute-aware Se- quence Network (ASN) to take the afore- mentioned users’ characteristics into account, which includes three modules: an attribute en- coder encodes the attribute preferences over the words; an attribute-aware review encoder adopts an attribute-based selective mechanism to select the important information of a re- view; and an attribute-aware summary de- coder incorporates attribute embedding and attribute-specific word-using habits into word prediction. To validate our model, we col- lect a new dataset TripAtt, comprising 495,440 attribute-review-summary triplets with three kinds of attribute information: gender, age, and travel status. Extensive experiments show that ASN achieves state-of-the-art performance on review summarization in both auto-metric ROUGE and human evaluation. 1 Introduction Review summarization aims to generate a con- densed summary for a review or multiple re- views1. Dominating studies can be divided into two groups: extractive and abstractive approaches. Extractive approaches (Hu and Liu, 2004; Gane- san, 2010) extract sentences or phrases from a review, while abstractive methods (Gerani et al., 1 Here, we focus on single-review summarization, and we leave adapting our model to multi-review summarization sce- nario to future work. Summary: trendy , stylish , but not meeting expectations R eview: definitely a night crowd type of hotel . . . very trendy common spaces , stylish. . and bold . however , the room felt a bit under done . . . too simple . location is great but very noisy even if we were on the top floor . will try another place next time . Attribute Information: traveled on business + businessperson preference for aspects + businessperson-specific vocabulary meeting, business, conference, internet, working Figure 1: A review-summary pair posted by a busi- nessperson in our dataset shows the effect of attribute information on summarizing review. Underlined words in the review indicate the important sentences that the businessperson care about when summarizing the re- view. Bold word in businessperson-specific vocabu- lary shows the businessperson’s word-using habits may help to generate the summary. 2014; Wang and Ling, 2016; Yang et al., 2018a; Li et al., 2019; …

Content truncated to 500 words