An Interactive Network for End-to-End Review Helpfulness Modeling
- PDF / 2,267,909 Bytes
- 19 Pages / 595.276 x 790.866 pts Page_size
- 49 Downloads / 170 Views
An Interactive Network for End‑to‑End Review Helpfulness Modeling Jiahua Du1 · Liping Zheng2 · Jiantao He3 · Jia Rong4 · Hua Wang1 · Yanchun Zhang1 Received: 9 March 2020 / Revised: 23 May 2020 / Accepted: 11 June 2020 © The Author(s) 2020
Abstract Review helpfulness prediction aims to prioritize online reviews by quality. Existing methods largely combine review texts and star ratings for helpfulness prediction. However, star ratings are used in a way that has either little representation capacity or limited interaction with review texts. As a result, rating information has yet to be fully exploited during the combination. This paper aims to overcome the two drawbacks. A deep interactive architecture is proposed to learn the text–rating interaction (TRI) for helpfulness modeling. TRI enlarges the representation capacity of star ratings while enhancing the influence of rating information on review texts. TRI is evaluated on six real-world domains of the Amazon 5-Core dataset. Extensive experiments demonstrate that TRI can better predict review helpfulness and beat the state of the art. Ablation studies and qualitative analysis are provided to further understand model behaviors and the learned parameters. Keywords Review helpfulness · Review texts · Star ratings · Text–rating interaction
1 Introduction Online reviews play an important role in the e-commerce ecosystem. Currently, online buyers highly rely on collective wisdom to make informed purchase decisions. A recent survey [43] shows that over 8 of 10 customers read reviews * Jiahua Du [email protected] Liping Zheng [email protected] Jiantao He [email protected] Jia Rong [email protected] Hua Wang [email protected] Yanchun Zhang [email protected] 1
Institute of Sustainable Industries and Liveable Cities, Victoria University, Melbourne, VIC, Australia
2
School of Computer Science and Technology, Fudan University, Shanghai, China
3
Guangzhou Metro Group Co., Ltd., Guangzhou, Guangdong, China
4
Faculty of Information Technology, Monash University, Clayton, VIC, Australia
for online retailers. The reviews also help manufactures collect user feedback and improve products. Nevertheless, the quality of user-generated reviews is uneven [34], susceptible to customers’ background, tolerance of product deficiencies, moods at the time of writing, to name a few. As the number of reviews grows, locating useful information becomes increasingly challenging. Many e-commerce platforms gather user voting on review helpfulness for quality assessment. Still, the voting data are scarce in practice and even missing in less popular products. Helpfulness prediction aims to identify and recommend high-quality reviews to customers in an automatic manner. The previous literature [2, 22, 44] largely employs review texts and star ratings for the task. The rationale lies in their ubiquitousness in contemporary online shopping platforms and their importance to review helpfulness modeling. Review texts qualitatively describe reviewers’ opinions toward produ
Data Loading...
