Product Question Answering in E-Commerce: A Survey

Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics Volume 1: Long Papers, pages 11951–11964 July 9-14, 2023 ©2023 Association for Computational Linguistics Product Question Answering in E-Commerce: A Survey Yang Deng1, Wenxuan Zhang2,†, Qian Yu3, Wai Lam1 1 The Chinese University of Hong Kong, 2 DAMO Academy, Alibaba Group, 3 JD.com {dengyang17dydy,isakzhang}@gmail.com, yuqian81@jd.com, wlam@se.cuhk.edu.hk Abstract Product question answering (PQA), aiming to automatically provide instant responses to cus- tomer’s questions in E-Commerce platforms, has drawn increasing attention in recent years. Compared with typical QA problems, PQA ex- hibits unique challenges such as the subjectivity and reliability of user-generated contents in E- commerce platforms. Therefore, various prob- lem settings and novel methods have been pro- posed to capture these special characteristics. In this paper, we aim to systematically review existing research efforts on PQA. Specifically, we categorize PQA studies into four problem settings in terms of the form of provided an- swers. We analyze the pros and cons, as well as present existing datasets and evaluation proto- cols for each setting. We further summarize the most significant challenges that characterize PQA from general QA applications and discuss their corresponding solutions. Finally, we con- clude this paper by providing the prospect on several future directions. 1 Introduction E-Commerce is playing an increasingly important role in our daily life. During the online shopping, potential customers inevitably have some questions about their interested products. To settle down their concerns and improve the shopping experi- ence, many AI conversational assistants have been developed to solve customers’ problems, such as Alexa (Carmel et al., 2018) and AliMe (Li et al., 2017a). The core machine learning problem under- lying them, namely Product Question Answering (PQA), thus receives extensive attention in both academia and industries recently. Figure 1 depicts an actual PQA example from Amazon. There are a ∗ The work described in this paper is substantially sup- ported by a grant from the Research Grant Council of the Hong Kong Special Administrative Region, China (Project Code: 14200719). † Corresponding author. Bikeroo Bike Seat Cushion-Comfortable Bicycle Saddle Brand Bikeroo Color Black Item Dimensions 10.2 x 7.8 x 2.36 inches Material Faux Leather Pad Type Waterproof Pad Product Info. Community QA Pairs Reviews R1: This cushion is high quality and very comfy. It holds up even if it gets wet. R2: I highly recommend this cushion. This seat cushion is comfortable and works well. Q1: what is the true size? A1: 10.2 wide (butt Bones) x 7.8 length x 2.4 high Question Is this really a good buy for cycling? Opinion YES Extraction highly recommend this cushion Retrieval This cushion is high quality and very comfy. It holds up even if it gets wet. Generation Yes, it is a very good seat cushion. It’s very comfortable and I love it. … … Figure 1: An PQA example from Amazon. tremendous amount of product-related data avail- able within the product page, which contains nat- ural language user-generated content (UGC) (e.g., product reviews, community QA pairs), structured product-related …

Content truncated to 500 words