Online Review Systems Research Articles

Win by Hook or Crook? Self-Injecting Favorable Online Reviews to Fight Adjacent Rivals

Literature has long assumed that unethical behaviors are fueled by competition. This study challenges this prevailing notion by introducing a conceptual distinction between rivalry and competition. We show that rivalry-induced deterrence triggers the formation of mutual forbearance in self-injecting favorable reviews. Mutual forbearance collapses when one party unilaterally engages in self-injecting behaviors, prompting the other party to retaliate by increasing their self-injecting activities. The presence of an additional rival leads to a 3.2% decrease in self-injecting intensity. Furthermore, self-injecting intensity increases by 0.095% for every 1% rise in that of the rivals and increases by 0.041% for a 1% rise in that of the nonrival competitors. Given the apparent role of rivalry in reducing self-injecting behaviors, platforms should strategically consider how system designs and policies can cultivate firms’ feelings of rivalry and avoid non-rival competition. We recommend platforms consider including evaluations from peer hotels as an alternative index for recommending hotels to potential consumers. Introducing peer evaluations to the current online review systems could attenuate the bias of consumer ratings and facilitate the formation of rivalry, which hinders self-injecting behaviors.

Comprehensive Sentiment and Fake Comment Analysis

This project introduces an innovative system for analyzing sentiment and detecting fake comments within e-commerce product reviews, aimed at enhancing the credibility and reliability of online feedback essential for consumer purchasing decisions. Our system integrates VADER, a robust rule-based model for general sentiment analysis, with RoBERTa, an advanced transformer-based deep learning model known for its superior accuracy. This dual approach harnesses the complementary strengths of traditional and modern techniques, resulting in a comprehensive and effective solution for sentiment analysis. A critical component of our system is the fake comment detection module, which utilizes sentiment scores and additional indicators to identify fraudulent reviews accurately. This module is essential for ensuring the authenticity of customer feedback and maintaining the integrity of online review platforms. Our analysis uncovers significant trends in sentiment across various product categories, providing valuable insights that can inform targeted marketing strategies and product enhancements. By examining the relationships between sentiment polarity, review ratings, and the likelihood of fake comments, we offer actionable intelligence for e-commerce platforms to improve product quality, enhance customer satisfaction, and safeguard their review systems. This project underscores the importance of advanced tools in upholding the trustworthiness of online reviews. By enabling businesses to understand genuine customer sentiment better and fostering transparency in the digital marketplace, our system enhances the overall consumer experience. Future efforts will focus on refining detection algorithms, exploring additional features for sentiment analysis, and validating the system across diverse datasets and languages to further elevate standards in online review systems.

Designing a deep learning-based application for detecting fake online reviews

In the era of prevalent online commerce, online reviews significantly influence purchasing decisions. Unfortunately, this has also led to the emergence of fake reviews, which can deceive consumers and undermine trust in online platforms. Our study addresses this issue by developing DenyBERT, a deep learning-based software that enhances the Bidirectional Encoder Representations from Transformers (BERT) framework with Deep and Light Transformation (DeLighT) and Knowledge Distillation (KD) techniques. These innovations not only reduce computational demands but also improve the model's accuracy in identifying fake reviews, making it ideally suited for real-world applications. Significantly, DenyBERT requires only 16.01M parameters-significantly fewer than its predecessors such as BERT and TinyBERT-yet achieves a robust accuracy of 96.12% and an F1-score of 96.47%. This efficiency makes it particularly suited for deployment on devices with limited processing capabilities. The software, developed in Python, features a flexible input mechanism that allows for analyzing reviews directly from websites via URL or via manual input of review paragraphs. Our findings indicate that DenyBERT outperforms existing models in both speed and accuracy, making it a powerful tool for combating fake reviews in real-time scenarios. This advancement not only enhances user trust in online review systems but also supports e-commerce platforms in maintaining a fair and transparent market environment.

Implementation of Multiview Deep Learning Algorithm using LSTM-CNN for Fraudulent Review Detection

In this project, we explored the task of detecting fraudulent reviews in text data using various machine learning algorithms. The dataset consisted of reviews categorized into two labels: 'CG' (genuine reviews) and 'OR' (potentially fraudulent reviews). We began by preprocessing the text data, which involved tasks such as removing punctuation, stop words, and applying stemming or lemmatization techniques. After preprocessing, we utilized different machine learning models including Logistic Regression, Decision Tree Classifier, Random Forest Classifier, Support Vector Machines (SVM), Multinomial Naive Bayes and LSTM-CNN model to classify the reviews. Overall, our study showcases the effectiveness of machine learning, Deep learning algorithms in identifying fraudulent reviews from textual data. The findings provide valuable insights for applications in e-commerce platforms and online review systems to enhance the reliability and trustworthiness of user-generated content. Keywords: Fraudulent reviews, Text classification, Machine Learning, Preprocessing, Logistic Regression, Support Vector Machines, Evaluation metrics.

Understanding Large-Scale Network Effects in Detecting Review Spammers

Opinion spam detection is a challenge for online review systems and social forum operators. Opinion spamming costs businesses and people money since it deceives customers as well as automated opinion mining and sentiment analysis systems by bestowing undeserved positive opinions on target firms and/or bestowing fake negative opinions on others. One popular detection approach is to model a review system as a network of users, products, and reviews, for example using review graph models. In this article, we study the effects of network scale on network-based review spammer detection models, specifically on the trust model and the SpammerRank model. We then evaluate both network models using two large publicly available review datasets, namely: the Amazon dataset (containing 6 million reviews by more than 2 million reviewers) and the UCSD dataset (containing over 82 million reviews by 21 million reviewers). It has been observed thatSpammerRank model provides a better scaling time for applications requiring reviewer indicators and in case of trust model distributions are flattening out indicating variance of reviews with respect to spamming. Detailed observations on the scaling effects of these models are reported in the result section.

The design of retailers' online review systems – a systematic literature review and future research agenda

PurposeIn this paper the authors aim to review the state-of-the-art literature on online review systems and their impacts on consumer behavior and retailers' performance with the aim of identifying research gaps related to different design features of review systems and developing future research agenda.Design/methodology/approachThe authors conducted a systematic review based on PRISMA 2020 protocol, focusing on studies published in the domains of retailing and marketing. This procedure resulted in 48 selected papers investigating the design features of retailer online review systems.FindingsThe authors identify eight design features that are controllable by retailers in an online review system. The design features have been researched independently in previous literature, with some features receiving more attention. Most selected studies focus on the design features adapted metrics and review presentations, while other features are generally neglected (e.g. rating dimensions). Previous literature argues that design features affect consumer behaviors and retailers' performance. However, the interactions among the features are still neglected in the literature, creating a relevant gap for future research.Originality/valueThis paper distinguishes between different types of retailer online review systems based on how they are implemented. The authors summarize the state-of-the-art of relevant literature on design features of online review systems and their effects on consumer- and retailer-related outcome variables. This systematic literature review distinguishes between online reviews provided on websites controlled by retailers (internal systems) and third-party websites (external systems).

“Sorry, too much information”—Designing online review systems that support information search and processing

When faced with a large number of reviews, customers can easily be overwhelmed by information overload. To address this problem, review systems have introduced design features aimed at improving the scanning, reading, and processing of online reviews. Though previous research has examined the effect of selected design features on information overload, a comprehensive and up-to-date overview of these features remains outstanding. We therefore develop and evaluate a taxonomy for information search and processing in online review systems. Based on a sample of 65 review systems, drawn from a variety of online platform environments, our taxonomy presents 50 distinct characteristics alongside the knowledge status quo of the features currently implemented. Our study enables both scholars and practitioners to better understand, compare and further analyze the (potential) effects that specific design features, and their combinations, have on information overload, and to use these features accordingly to improve online review systems for consumers.

MOOC Data Analytics through Text Mining-An Innovative Approach to Learning Improvement

Abstract: The COVID-19 pandemic has brought about significant changes in the perception of education, with Massive Open Online Course (MOOC) providers like Coursera witnessing a surge in millions of new user registrations on their platforms. However, despite the prevalence of online review systems in various industries, the MOOC ecosystem lacks a standardized or fully decentralized review system. We believe that there is an opportunity to utilize existing open MOOC reviews to create userfriendly and transparent reviewing systems, enabling learners to easily identify the top courses available. By leveraging the wealth of reviews already available in the MOOC ecosystem, we can create simpler and more transparent systems that empower users to make informed choices about the courses they enrol in. In our research, we conduct an analysis of reviews from the Coursera platform with the specific goal of determining the potential value of using NLP-driven sentiment analysis on textual reviews in providing valuable information to learners. By examining the sentiment expressed in the textual reviews, we aim to evaluate whether this approach can offer meaningful insights to learners in assessing the quality of MOOCs. The results of our research suggest that textual reviews may be a more advantageous choice compared to numeric ratings due to the disadvantages associated with numeric ratings, such as the potential for random or arbitrary selections. Our findings indicate that utilizing sentiment analysis on textual reviews could provide valuable information for learners in evaluating the quality of MOOCs. By relying on the rich and descriptive information conveyed through textual reviews, learners may be better equipped to make informed decisions when selecting courses on platforms like Coursera.

Towards a Review-Analytics-as-a-Service (RAaaS) Framework for SMEs: A Case Study on Review Fraud Detection and Understanding

With the advancement of internet technology, customers increasingly rely on online reviews as a valuable source of information. The study aims to develop a marketing data analytics framework to manage online reviews, especially fake reviews, which have become a significant issue undermining the creditability of online review systems. As small and medium-sized enterprises often lack the capabilities to automatically derive customer insights from online reviews, this study proposes a cost-effective, extensible Review-Analytics-as-a-Service (RAaaS) framework that can be operated by non-data specialists to facilitate online review data analytics. We demonstrate the framework’s application by using two datasets with more than 400,000 online reviews from Yelp to simulate live platforms and demonstrate an analytic flow of review fraud detection and understanding. The findings reveal insights into the influence of fake reviews on product ranking and exposure rate. Moreover, it was found that there was a higher concentration of sadness and anger in fake reviews (vs. organic reviews). In addition, fake reviews tend to be shorter, more extreme (with the use of strong adverbs), and have different patterns of topic distribution. This study has important implications for different stakeholder groups including, but not limited to, SMEs, review platforms and customers.

A Warning Approach to Mitigating Bandwagon Bias in Online Ratings: Theoretical Analysis and Experimental Investigations

Current online review systems widely suffer from rating biases. Biased ratings can lead to violations of customer trust and failures of business intelligence. Hence, both practitioners and researchers have directed massive efforts toward curbing rating biases. In this paper, we investigate bandwagon bias, the rating distortion resulting from individuals posting ratings shifted toward the displayed average rating, and propose a bias warning approach to mitigate this bias. Drawing on the flexible correction model, the theory of valuation in behavioral economics, and previous warning research, we design an effective warning strategy in two steps. First, we start with the risk-alert warning strategy, which prior research has widely employed, and rationalize its deficiencies by synthesizing theoretical analysis and extant empirical evidence. Second, considering the deficiencies, we identify a supplementary content design factor—the ranking task—and construct a risk-alert-with-ranking-task warning strategy. We then empirically test the effects of the two warning strategies on individual ratings in cases in which bandwagon bias either occurs or does not occur in individuals’ initial assessments. The results of four controlled experiments indicate that (1) the risk-alert strategy can reduce bandwagon bias in individual ratings but will elicit unwanted rating distortions when bandwagon bias does not occur in individuals’ initial assessments, and (2) the risk-alert-with-ranking-task strategy can mitigate bandwagon bias while avoiding the unwanted rating distortions above and can thus function as an effective warning strategy. Our research contributes to the literature by proposing an effective debiasing solution for bandwagon bias and a bias warning approach for online rating debiasing, which can help increase rating informativeness on online platforms.

What makes an online review credible? A systematic review of the literature and future research directions

Online reviews of products and services are strategic tools for e-commerce platforms, as they aid in consumers’ pre-purchase decisions. Past research studies indicate online reviews impact brand image and consumer behaviour. With several instances of fake reviews and review manipulations, review credibility has become a concern for consumers and service providers. In recent years, due to growing webcare attitude among managers, the need for maintaining credible online reviews on the e-commerce platforms has gained attention. Though, there are several empirical studies on review credibility, the findings are diverse and contradicting. Therefore, in this paper, we systematically review the literature to provide a holistic view of antecedents of online review credibility. We examine variables, methods, and theoretical perspective of online review credibility research using 69 empirical research papers shortlisted through multi-stage selection process. We identify five broad groups of antecedents: source characteristics, review characteristics, consumer characteristics, interpersonal determinants in the social media platform and product type. Further, we identify research issues and propose directions for future research. This study contributes to existing knowledge in management research by providing the holistic understanding of the “online review credibility” construct and helps understand what factors lead to consumers’ belief in the credibility of online review. The insights gained would provide managers adequate cues to design effective online review systems.

Online Patient Reviews of Breast Reconstruction: RealSelf Analysis.

The breast reconstruction category from RealSelf.com was analyzed using a web crawler-based application built from Python and Selenium. Reviews were collected from May 2009 to November 2021. Information including RealSelf's inherent "worth it" ranking system, review text, the number of submitted photographs, and the number of readers who found the review helpful was captured. The content of the review was then independently reviewed by the authors and was categorized with key factors that determined positive or negative reviews. A total of 3451 breast reconstruction reviews were collected. After the authors analyzed each review, 3225 (94.33%) were identified as positive reviews. The most common factors associated with positive reviews were physician demeanor (n = 2600, 31.7%), aesthetic outcome (n = 1955, 23.8%), or staff (n = 1543, 18.8%), while negative reviews were associated with unfavorable aesthetic outcome (n = 94, 28.9%), physician demeanor (n = 82, 25.2%), or postoperative complications (n = 75, 23.1%). Although there are surveys that analyze patient satisfaction for breast reconstruction, there has not been a study that analyzed a large online review database. Predominating factors in both positive and negative reviews were physician demeanor and aesthetic outcome.

Design is More Than Looks: Research on the Affordance of Review Components on Consumer Loyalty.

Online review system contains multiple components, such as ratings, review text, product pictures, and video uploads, that could affect consumer loyalty. However, how the affordance of such components influences perceptions and behaviors of consumers remains unclear. We extend stimulus-organism-response (S-O-R) theory to the online review system. Specifically, we combine affordance theory and the technology acceptance model (TAM) to investigate the relations among the affordance of review systems, consumers' perceived beliefs, and their loyalty. We surveyed 320 customers on their online shopping experiences in China. We tested our hypotheses using the partial least squares path structural equation modeling (PLS-SEM) method. We report the direct effect of affordances of review components on consumer loyalty and its indirect effects on consumer loyalty through perceived beliefs. Our results show that integrity and social interaction affordance of review components have significant relations with perceived ease of use, perceived usefulness, and perceived enjoyment. Intelligent topic mining reveals a positive relation on perceived enjoyment. Operability has a positive relation with perceived ease of use and perceived usefulness. These three consumer-perceived beliefs can mediate, to different degrees, the relationship between affordance of review components and consumer loyalty. This research takes an innovative approach to offer insights into the relationships between IT affordances and consumer perceptions. We examine S-O-R theory through the lens of information technology and extend S-O-R theory by integrating IT affordances. Our research findings pave the way for businesses to design and implement more effective online review systems.

Multicultural Ability in Dental English Impressions of Hypermedia Email: A Specific Educational Paradigm

Given the noteworthy connection between special English education acuities and grade point average (GPA), as a directory of educational success, and settlement, it is essential to support the students’ English education talents. Two-way ANCOVA was engaged to evaluate the knowledge efficiency of the informal AI syllabus with dissimilar learning method (i.e., the series of perform schemes and the set of empirical learning) and masculinity. A cross-sectional appraisal was operated through online review and fractional slightest systems was applied in evaluating the records. Universal English for Dental Purposes (EDP) is educated from the opinion of actual conditions and it implies that education English is absorbed on a specific texts and particular terminology. But what is more significant education EDP terminology trusts on emerging statement services, difficulty-resolving, and choice production (Milosavljevic et al., 2015). The investigation recommends that scholars are of the opinion that more communication is continued through e-mail than might have been possible in communication interaction and more, that the English knowledge helps as valuable framework for more acquiring. On the foundation of the contributors' support and reaction, the probable educational worth of an e-mail may converse action in EDP program is undisputable.
 The placement-plan reaction provides durable indication of the possible educational profits of the e-mail scheme. Academic communication through e-mail replace to provide a real method to connect the opening between teaching EDP happenings and the energetic firm-changing education setting.

Can Customers Affect Racial Discrimination in Hiring?

While significant scholarship has documented the prevalence of racial discrimination in hiring, less is known about the forces that exacerbate or mitigate it. In this article, we develop a theoretical argument about the ability of customers to influence racial discrimination in hiring, highlighting the role of direct customer communication and its intersection with online review systems. We deploy a novel method to test our argument. Specifically, we draw on original data from a two-part field experiment that first randomly assigned restaurants to receive one of three different email messages from customers and then audited the restaurants to test for racial discrimination in hiring. While our data collection effort was cut short and disrupted by the coronavirus pandemic, making our findings more exploratory than initially anticipated, our data provide evidence that customer communication can reduce racial discrimination under certain conditions. We discuss the implications of these findings for scholarship on organizational decision-making, discrimination, and methodological approaches for studying these topics.

Explaining reviewing effort: Existing reviews as potential driver

Online review systems try to motivate reviewers to invest effort in writing reviews, as their success crucially depends on the helpfulness of such reviews. Underlying cognitive mechanisms, however, might influence future reviewing effort. Accordingly, in this study, we analyze whether existing reviews matter for future textual reviews. From analyzing a dataset from Google Maps covering 40 sights across Europe with over 37,000 reviews, we find that textual reviewing effort, as measured by the propensity to write an optional textual review and (textual) review length, is negatively related to the number of existing reviews. However, and against our expectations, reviewers do not increase textual reviewing effort if there is a large discrepancy between the existing rating valence and their own rating. We validate our findings using additional review data from Yelp. This work provides important implications for online platforms with review systems, as the presentation of review metrics matters for future textual reviewing effort.

Large scale analysis of open MOOC reviews to support learners’ course selection

The recent pandemic has changed the way we see education. During recent years, Massive Open Online Course (MOOC) providers, such as Coursera or edX, are reporting millions of new users signing up on their platforms. Though online review systems are standard among many verticals, no standardized or fully decentralized review systems exist in the MOOC ecosystem. In this vein, we believe that there is an opportunity to leverage available open MOOC reviews in order to build simpler and more transparent reviewing systems, allowing users to really identify the best courses out there. Specifically, in our research we analyze 2.4 million reviews (which is the largest MOOC reviews dataset used until now) from five different platforms in order to determine the following: (1) if the numeric ratings provide discriminant information to learners, (2) if NLP-driven sentiment analysis on textual reviews could provide valuable information to learners, (3) if we can leverage NLP-driven topic finding techniques to infer themes that could be important for learners, and (4) if we can use these models to effectively characterize MOOCs based on the open reviews. Results show that numeric ratings are clearly biased (63% of them are 5-star ratings), and the topic modeling reveals some interesting topics related with course advertisements, the real applicability, or the difficulty of the different courses.

Fake Review Detection on Using Machine Learning on Online Product Selling Platform

Abstract: The increasing popularity of online review systems motivates malevolent intent in competing sellers and service providers to manipulate consumers by fabricating product/service reviews. Immoral actors use Sybil accounts, Bot farms, and purchase authentic accounts to promote products and vilify competitors. Facing the continuous advancement of review spamming techniques, more research work is been carried out to assess the approaches explored to date to combat fake reviews, and regroup to define new ones. Fake reviews detection attracts many researchers’ attention due to the negative impacts on the society. Most existing fake reviews detection approaches mainly focus on semantic analysis of review’s contents. This project is aimed at fake review detection in online platform, to prevent damage due to deceptive reviews.We propose a Novel Fake Reviews Detection based on Logistic Regression technique

C2C interaction in the online review system: Effect of other customers’ responses on subsequent review volume

Customer-to-customer (C2C) interactions can be realised through other customers’ responses (OCRs) in the online review system. However, the externality of such interactions has not been clearly understood. We address this deficiency by exploring the effects of OCRs on subsequent review volume. Drawing on social learning theory and reinforcement theory, we hypothesise that OCRs have positive effects on subsequent review volume through the visibility mechanism. We verify this hypothesis by matching restaurants across two online platforms using a difference-in-differences strategy. The results show that restaurants with OCRs show an average increase of 5.286 reviews per month, but the effect fades if subsequent customers have not observed OCRs in the last five review pages. We further validate the visibility mechanism and conduct robustness checks by examining review heterogeneity and performing within-platform analysis. Our findings highlight the importance of C2C interactions and provide implications for the design and management of online review systems.

An Empirical Study on the Differences between Online Picture Reviews and Text Reviews

In the context of e-commerce, online travel agencies often derive useful information from online reviews to improve transactions. Based on the dispute on the usefulness of different types of reviews and social exchange theory, this study investigates how the characteristics of pictures and text influence review reading and review posting behaviors and thus influencing the efficiency of online review systems. By analyzing crawled data of online hotels and conducting experiments, we first find that picture reviews are more useful than text reviews, and high-quality pictures in reviews have a significant impact on review usefulness. Second, posting pictures requires review posters to pay more perceived costs. Third, negative review posters have higher perceived costs, so they are more unwilling to post pictures, especially high-quality pictures. Our results indicate that review platforms need to add incentives to encourage consumers to post high-quality picture reviews and design workable interfaces to reduce the burden of negative reviewers to speed up the purchase decision process for review readers. This study provides theoretical implications by demonstrating how the adoption of the picture in review systems influences both review readers’ and review posters’ behaviors. Additionally, our findings also provide useful managerial insights for online travel suppliers in terms of building an effective review system to promote sales.