Multi-perspective Coherent Reasoning for Helpfulness Prediction of Multimodal Reviews

Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, pages 5927–5936 August 1–6, 2021. ©2021 Association for Computational Linguistics 5927 Multi-perspective Coherent Reasoning for Helpfulness Prediction of Multimodal Reviews Junhao Liu1,2,3∗ Zhen Hai3 Min Yang1† Lidong Bing3 1Shenzhen Key Laboratory for High Performance Data Mining, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences 2University of Chinese Academy of Sciences 3DAMO Academy, Alibaba Group {jh.liu, min.yang}@siat.ac.cn {zhen.hai, l.bing}@alibaba-inc.com Abstract As more and more product reviews are posted in both text and images, Multimodal Review Analysis (MRA) becomes an attractive re- search topic. Among the existing review analy- sis tasks, helpfulness prediction on review text has become predominant due to its importance for e-commerce platforms and online shops, i.e. helping customers quickly acquire use- ful product information. This paper proposes a new task Multimodal Review Helpfulness Prediction (MRHP) aiming to analyze the re- view helpfulness from text and visual modal- ities. Meanwhile, a novel Multi-perspective Coherent Reasoning method (MCR) is pro- posed to solve the MRHP task, which con- ducts joint reasoning over texts and images from both the product and the review, and ag- gregates the signals to predict the review help- fulness. Concretely, we first propose a product- review coherent reasoning module to measure the intra- and inter-modal coherence between the target product and the review. In addi- tion, we also devise an intra-review coherent reasoning module to identify the coherence between the text content and images of the review, which is a piece of strong evidence for review helpfulness prediction. To evalu- ate the effectiveness of MCR, we present two newly collected multimodal review datasets as benchmark evaluation resources for the MRHP task. Experimental results show that our MCR method can lead to a performance increase of up to 8.5% as compared to the best performing text-only model. The source code and datasets can be obtained from https:// github.com/jhliu17/MCR. 1 Introduction Product reviews are essential information sources for consumers to acquire useful information and ∗This work was conducted when Junhao Liu was an intern at DAMO Academy, Alibaba Group. †Min Yang is the corresponding author. make purchase decisions. Many e-commerce sites such as Amazon.com offer reviewing functions that encourage consumers to share their opinions and experiences. However, the user-generated reviews vary a lot in their qualities, and we are continuously bombarded with ever-growing, noise information. Therefore, it is critical to examine the quality of reviews and present consumers with useful reviews. Motivated by the demand of gleaning insights from such valuable data, review helpfulness pre- diction has gained increasing interest from both academia and industry communities. Earlier re- view helpfulness prediction methods rely on a wide range of handcrafted features, such as semantic fea- tures (Yang et al., 2015), lexical features (Martin and Pu, 2014), and argument based features (Liu et al., 2017), to train a classifier. The success of these methods generally relies heavily on feature engineering which is labor-intensive and highlights the weakness …

Content truncated to 500 words