Creating and detecting fake reviews of online products

Customers increasingly rely on reviews for product information. However, the usefulness of online reviews is impeded by fake reviews that give an untruthful picture of product quality. Therefore, detection of fake reviews is needed. Unfortunately, so far, automatic detection has only had partial success in this challenging task. In this research, we address the creation and detection of fake reviews. First, we experiment with two language models, ULMFiT and GPT-2, to generate fake product reviews based on an Amazon e-commerce dataset. Using the better model, GPT-2, we create a dataset for a classification task of fake review detection. We show that a machine classifier can accomplish this goal near-perfectly, whereas human raters exhibit significantly lower accuracy and agreement than the tested algorithms. The model was also effective on detected human generated fake reviews. The results imply that, while fake review detection is challenging for humans, “machines can fight machines” in the task of detecting fake reviews. Our findings have implications for consumer protection, defense of firms from unfair competition, and responsibility of review platforms.     

How online reviews and coupons affect sales and pricing: An empirical study based on e-commerce platform

With rapid economic growth largely fueled by digital technologies, online reviews are among the new wave of technologies that still make a significant contribution to this new digital economy. However, there is scarce academic research in promoting or responding online reviews using reduction coupons with threshold issued by sellers and pricing decisions. We employ a three-stage least squares (3SLS) model to estimate the effects of online reviews and coupons on online product sales and price. We also extend our research using the data of two product categories. Our results mainly include: (1) Negative impact of negative online reviews on sales is moderated by price, and consumers are more tolerant of negative reviews of high-priced products. (2) Consumers' perceived usefulness of online coupons helps promote more sales and weaken the relationship between negative reviews and sales. (3) Positive online reviews weaken the negative relationship between online negative reviews and the price of product. (4) Consumers’ increased perceived usefulness of online coupons will encourage sellers to set a higher price and weaken the negative relationship between negative reviews and the price of product. (5) Compared with the search product, sellers regard negative reviews from the product with experience attributes as less helpful and apply less effective coupon promotion to response to negative reviews. Our study clarifies the interaction between online reviews and online coupons on e-commerce platforms under the operation scenario. We also provide empirical support for accurate promotion and higher sales revenue through reasonable pricing decisions and specified reduction coupons.     

基于LDA模型及情感分析的电影主题公园评论文本研究

以中国电影主题公园为例，收集多个旅游网站游客评论数据，采用LDA主题模型方法、共现网络分析方法及情感分析方法对游客评论文本进行研究。结果表明，国内电影主题公园游客评论可以分为旅游服务主题、旅游景观主题、整体感知主题以及活动体验主题4个主题。结合共现网络分析及情感分析结果识别当前景区现状及痛点，为文旅企业经营者提供较为准确的游客偏好信息，也为企业后续管理及决策提供相关理论及数据支撑。     

The informational value of multi-attribute online consumer reviews: A text mining approach

In the digital age, customers use online reviews to minimize the risks associated with purchasing a product. Major online retailers help customers choose the right product by exposing reviews that received many “helpful” votes at the top of the review section. Given that reviews that have received the maximum helpfulness votes are considered more important in purchase decisions, understanding determinants of helpfulness votes offers clear benefits to online retailers and review platforms. This study focuses on the effect of review informativeness, which is measured by the number of attributes discussed in a review, and its interplay of review valence on customers' perception of review helpfulness. We applied a word-level bigram analysis to derive product attributes from review text and examined the influence of the number of attributes on the review's helpfulness votes. More importantly, we also suggested the moderating role of review valence. Estimation results of the Zero-inflated Poisson models on 21,125 reviews across 14 wireless earbuds indicated that as more attributes are discussed in a review, the more the review can earn helpfulness votes from customers. Furthermore, the positive association between the number of attributes and helpfulness was enhanced among negative reviews. This study contributes to customers' information processing literature and offers guidelines to online retailers in designing a better decision support system.     

Addressing negative product attribute reviews: The case of advertising product attributes and corporate social responsibility information

Negative product reviews are bad for businesses. They can adversely affect product sales, brand evaluations, and brand loyalty. To attend to negative reviews, one approach is through advertising products or company virtues. The current study examines the efficacy of different advertising approaches (core product attribute advertising vs. non-core product attribute advertising vs. corporate social responsibility [CSR] advertising) in helping dilute the negative effect of dissimilar negative product attribute reviews (negative core product attribute reviews vs. negative non-core product attribute reviews). Through four experiments, we found that when the non-core attribute of a product had negative reviews, a core attribute advertising approach weakened the effect of negative reviews. However, CSR advertising was less effective. Furthermore, when the core attributes of a product had negative reviews, a non-core attribute advertising effort did not weaken, but actually augmented the impact of negative reviews. In addition, when the core attributes of a product received negative reviews, CSR advertising decreased the influence of negative reviews. Our findings provide theoretical and managerial implications for managing negative online reviews of product attributes.     

Emotional and the normative aspects of customers’ reviews

The goal of the current research is to investigate the link between the emotional aspects of hotel and travel organization customers' reviews and their normative (e.g., star rating) rankings. After filtering, the Yelp dataset generated 3,47,803 hotel and travel company reviews. Following the purification of user reviews, we used an unsupervised machine learning technique-based NRC Emotion Lexicon to study the relationships between various emotional aspects of reviews and their normative values (e.g., star rating) for the review. Customers express different sorts of feelings for different types of emotional aspects, forcing them to assign different stars, according to the study's findings. The study is the first to use a lexicon-based unsupervised learning approach to look into the emotional aspects of hotel and travel organization reviews and associated normative (e.g., star rating) rankings.     

Clustering and compatibility between tourism attractions

Cumulative attraction or compatibility is a powerful factor in locational decisions of retail outlets, but it has received little attention in the tourism literature. Existing studies largely ignore the impact of spatial and thematic clustering on compatibility between tourism businesses. This research questions whether and how spatial and thematic clustering are related to tourist movements between visitor attractions. The study in Cornwall, England, is based on in‐depth interviews with tourist attraction managers and key informants, complemented by a survey of 435 tourists. It provides insights into the relationship between the nature of the tourism product, spatial clustering and tourism behaviour. Copyright © 2009 John Wiley & Sons, Ltd.     

The effectiveness of environmental interpretation at resource‐Sensitive tourism destinations

The process of appropriate visitor management is an integral part of sound tourism management. Visitor management includes the development and implementation of rules and regulations with respect to visitor activity, which in turn provides the guidelines for visitors. On the other hand, it also aims to create enjoyable visitor experiences in the hope that visitors will appreciate the value of the site. Environmental interpretation in a resource‐sensitive tourism destination is considered to be an effective visitor management strategy that helps to encourage visitors to adopt more appropriate behaviour in order to sustain the development of tourism. This paper aims to examine the processes and purposes of visitor management and environmental interpretation, including relevant definitions and functions. The characteristics of visitor management and environmental interpretation, and how to increase their effectiveness, also are addressed. Copyright © 2002 John Wiley & Sons, Ltd.