Content uploaded by Panos Ipeirotis
Author content
All content in this area was uploaded by Panos Ipeirotis
Content may be subject to copyright.
A preview of the PDF is not available
Designing Ranking Systems for Consumer Reviews: The Impact of Review Subjectivity on Product Sales and Review Quality
Abstract
With the rapid growth of the Internet, users' ability to publish content has created active electronic com-munities that provide a wealth of product information. Consumers naturally gravitate to reading reviews in order to decide whether to buy a product. However, the high volume of reviews that are typically pub-lished for a single product makes it harder for individuals to locate the best reviews and understand the true underlying quality of a product based on the reviews. Similarly, the manufacturer of a product wants to identify the reviews that influence the customer base, and examine the content of these reviews. In this paper we propose two ranking mechanisms for ranking product reviews: a consumer-oriented ranking mechanism ranks the reviews according to their expected helpfulness, and a manufacturer-oriented rank-ing mechanism ranks the reviews according to their expected effect on sales. Our ranking mechanism combines econometric analysis with text mining techniques in general, with subjectivity analysis in par-ticular. We show that subjectivity analysis can give useful clues about the helpfulness of a review and about its impact on sales. Our results can have several implications for the market design of online opin-ion forums.
... Further, the Bright local's survey in 2020 found that 87% of consumer read products reviews and the time average that spend reading is 13 min and 45sec [4]. Previous researches have also caught attention in electronic word-of-mouth (eWOM), since it has a positive increment on online sales [5][6][7].Moreover, the impact of covid-19 in e-commerce has shown a peak on sales in online business from 10% to 50 %, and is expected to grow more in the following years. Therefore, ORs as a potential source of social influence, will be continuous analyze in the upcoming years. ...
... After the data collection, ORs with less than 3 Helpful votes were eliminated from the analysis because as previous researches [6] and [44], authors argued that the analysis must consider reviews that assure number of votes. ...
Online reviews (ORs) have shown evidence to help consumers to reduce hesitation in the last stage of the purchase and has been also found that ORs help online businesses increase sales. However, ORs are increasing faster, becoming every time more robust with better ways and media to express useful and helpful information. Therefore, the way ORs help online business and consumers are constantly changing. Previous studies have intended to analyze helpfulness in different ways. However, they have not totally yet identified the most appropriate influence significance of the factors to test and predict the helpfulness of ORs due to the constant change and evolution of ORs in E-commerce platforms. I based this study on the economics of information, media richness, and negativity-bias theories, proposing a model that shows the influencing factors in the helpfulness of ORs (such as length, sentimental Analysis, score rating, number of images, video and published days). To find a closer optimal helpfulness analysis and prediction, a data set of 17,119 samples of three types of online goods have been extracted from different products on Amazon.com. For the analysis, we have considered employing a regression model to analyze the significance level of the factors in ORs for every type of online goods. The findings in this research prove that in fact there is a different perception of helpfulness for every type of good.
... El número de opiniones sobre cada producto o servicio ofertado en una web puede ser utilizado como una aproximación al número de unidades vendidas, como se ha hecho en estudios previos con Amazon (Chevalier and Mayzlin, 2006;Ghose and Ipeirotis, 2007). En el sector turístico también se ha utilizado esta técnica utilizando datos de Booking.com ...
... El número de opiniones sobre cada producto o servicio ofertado en una web puede ser utilizado como una aproximación al número de unidades vendidas, como se ha hecho en estudios previos con Amazon (Chevalier and Mayzlin, 2006;Ghose and Ipeirotis, 2007). En el sector turístico también se ha utilizado esta técnica utilizando datos de Booking.com ...
... IMDb, eBay, and many others. There is, however, a high level of subjectivity in all of the aforementioned rating systems (Ghose and Ipeirotis 2006). A simple rating system, such as the unary or binary ratings, fails to provide sufficient granularity for distinguishing products from each other, whereas a system that uses a wide rating scale, such as the 100-point rating scale, is often subject to misinterpretation of the rating scores. ...
With an increasing number of product reviews available online, it has become impractical for potential customers to perceive all the available reviews in order to make an informed decision on their purchase. Product ratings that encapsulate product reviews swiftly and easily have become an alternative for customers. However, since several product ratings only display the overall rating, customers may still find it challenging to make an informed decision due to the lack of information between positive and negative reviews. In addition, existing product ratings are static in nature as they do not cater to customers’ different needs since they often prioritize different aspects of the product or product features. Accordingly, this paper proposes a dynamic aspect-based rating system accompanied by an aspect-based rating visualization to address the aforementioned problems. This rating system also considers the users’ reputations who have given product reviews to give a more holistic view of the users posting reviews. Moreover, our user study shows that our proposed rating visualization can be a competitive alternative in representing a product rating since it has the advantage of being informative and easily customized due to its ability to display rating scores based on users’ preferred aspects. In addition, the proposed visualization also enables customers to make more informed decisions since it displays a balance of both positive and negative reviews.
Live streaming has enabled viewers to interact with broadcasters and peer viewers, as well as tip the broadcaster based on their emotional state. While most tips are small, a minority of viewers tip extremely large. To uncover such unique tipping patterns, we draw on the emotion literature and competitive arousal theory to examine how viewers’ emotional extremity affects tipping behavior and how broadcaster's emotional response moderates the baseline effect. This study contributes to the emerging live streaming literature by revealing the differential effects of emotional extremity on tipping amount and frequency and provides practical implications for broadcasters and platforms to steer tipping behavior.
Sentiment analysis is also known as opinion mining. It is one of the critical and rapidly growing fields in Natural Language Processing. This research is intended to address the most fundamental difficulties relating to sentiment analysis. Sentiment polarity classification. A generic technique for categorizing sentiment polarity, along with detailed processes, is presented. The input data for this project is taken from major e-commerce websites like Amazon’s product ratings. Analyses on sentence-part classification and review-part categorization have shown promising results. At last, insights will be derived from preprocessed data for future execution and how much product structure and features impact execution. Created a significant impact on improving the business by producing a quality product, gadget, or accessory that customers expect and satisfying their expectations.
Online reviews have an undeniable impact on the market and are an important source of consumer information. From a legal perspective, online reviews actively influence consumers’ decisions to enter into a contract. Moreover, online reviews convey pre-contractual information that consumers find relevant and easy to understand, unlike the pre-contractual information disclosed as a result of EU law–based information duties. From this perspective, online reviews could potentially be seen as a complement of the flawed EU law–based information paradigm and regulatory improvement options based on reviews could be explored. However, the unreliability of online reviews is an obstacle that haunts consumers, practitioners, regulators, and academics alike. This unreliability has previously been identified as a reason not to award online reviews a more significant role in the EU law–based regulatory framework of pre-contractual information in consumer contracts. This paper explores the merits of this argument by discussing how the unreliability of online reviews is currently regulated. This paper takes a broad perspective on regulation, focusing not only on EU consumer legislation, but also looking at standardization, soft law, self-regulation, and the role of national consumer authorities. Overall, this paper argues that there are sufficient measures in place to shift the debate from the unreliability of reviews to reviews’ potential role in the protection of consumer informational interests.
Employer review sites have grown popular over the last few years, with 86 percent of job seekers referring to reviews on these sites before applying to job positions. Though the antecedents of review helpfulness have been studied in various contexts, it has received limited attention in the employee review context. These sites provide review text in multiple dimensions, such as pros and cons. Besides, to solicit unbiased reviews, these sites allow an option of keeping reviewer information anonymous. Rooted in the diagnosticity perspective, we investigate review helpfulness focusing on the role of review text in multiple dimensions and the anonymity of the reviewers. We use a publicly available Glassdoor dataset to model review helpfulness using a Tobit regression. The results show that the review length in multiple dimensions of review text and anonymity positively impact review helpfulness. Moreover, anonymity positively moderates the review length in the cons section. As a post-hoc analysis, we perform topic modeling to gain better insights on the review text in multiple dimensions and anonymity. The post-hoc analyses show that non-anonymous reviewers discuss firm reputation in the pros section, which anonymous reviewers do not. In the cons section, non-anonymous reviewers discuss politics, unfair and unethical treatment, and prospects of the employer, while anonymous reviewers discuss incompetency of the leadership. This research has important practical implications for online review sites’ design and crafting guidelines and policies for employees writing reviews.
With the rapid development of Internet economy, e-commerce websites have accumulated a large number of users’ online comment data, which is an important way for users to express their real needs and emotions. The survey takes the consumer reviews of the elderly wearable devices as the research object, conducts in-depth mining and emotional analysis on the review data and feature words, and obtains the characteristics and functions of wearable devices that consumers care about, as well as the main factors influencing consumers to make purchase decisions.
This paper presents a simple unsupervised learning algorithm for classifying reviews as recommended (thumbs up) or not rec- ommended (thumbs down). The classifi- cation of a review is predicted by the average semantic orientation of the phrases in the review that contain adjec- tives or adverbs. A phrase has a positive semantic orientation when it has good as- sociations (e.g., "subtle nuances") and a negative semantic orientation when it has bad associations (e.g., "very cavalier"). In this paper, the semantic orientation of a phrase is calculated as the mutual infor- mation between the given phrase and the word "excellent" minus the mutual information between the given phrase and the word "poor". A review is classified as recommended if the average semantic ori- entation of its phrases is positive. The al- gorithm achieves an average accuracy of 74% when evaluated on 410 reviews from Epinions, sampled from four different domains (reviews of automobiles, banks, movies, and travel destinations). The ac- curacy ranges from 84% for automobile reviews to 66% for movie reviews.
Three of the most important uses of the Internet today are as an economic marketplace, as aforum for social interaction, and as a source of information. In this paper, we explore how thesethree activities come together, in the form of emergent social communities built around informationexchanges within IT-enabled electronic marketplaces. Drawing on social identity theory, we suggestthat the relationship between online consumer reviews and internet product sales is partiallyexplained by social identity processes. Using a unique dataset based on both chronologicallycompiled ratings as well as reviewer characteristics for a given set of products and geographicallocation-based purchasing behavior from Amazon, we provide evidence at the community levellinking the prevalence of identity claiming behavior in an online community with subsequentproduct sales. In addition, we show that when reviewers claim to be from a particular geographiclocation, subsequent product sales are higher in that region. At the review level of analysis, we showthat subsequent reviews conform to identity-claiming norms set in previous reviews, and thatidentity claiming that conforms to community norms elicits identity granting. Furthermore, ourresults suggest that the prevalence of identity granting has implications for economic exchange inthe form of product sales. Implications for research on word-of-mouth and electronic communitiesare discussed.
Identifying sentiments (the affective parts of opinions) is a challenging problem. We present a system that, given a topic, automatically finds the people who hold opinions about that topic and the sentiment of each opinion. The system contains a module for determining word sentiment and another for combining sentiments within a sentence. We experiment with various models of classifying and combining sentiment at word and sentence levels, with promising results.
Merchants selling products on the Web often ask their customers to review the products that they have purchased and the associated services. As e-commerce is becoming more and more popular, the number of customer reviews that a product receives grows rapidly. For a popular product, the number of reviews can be in hundreds or even thousands. This makes it difficult for a potential customer to read them to make an informed decision on whether to purchase the product. It also makes it difficult for the manufacturer of the product to keep track and to manage customer opinions. For the manufacturer, there are additional difficulties because many merchant sites may sell the same product and the manufacturer normally produces many kinds of products. In this research, we aim to mine and to summarize all the customer reviews of a product. This summarization task is different from traditional text summarization because we only mine the features of the product on which the customers have expressed their opinions and whether the opinions are positive or negative. We do not summarize the reviews by selecting a subset or rewrite some of the original sentences from the reviews to capture the main points as in the classic text summarization. Our task is performed in three steps: (1) mining product features that have been commented on by customers; (2) identifying opinion sentences in each review and deciding whether each opinion sentence is positive or negative; (3) summarizing the results. This paper proposes several novel techniques to perform these tasks. Our experimental results using reviews of a number of products sold online demonstrate the effectiveness of the techniques.
Sciences Workshop for useful suggestions and comments. Please send all correspondence on this manuscript to: Thomas Y.
Sentiment analysis seeks to identify the view- point(s) underlying a text span; an example appli- cation is classifying a movie review as "thumbs up" or "thumbs down". To determine this sentiment po- larity, we propose a novel machine-learning method that applies text-categorization techniques to just the subjective portions of the document. Extracting these portions can be implemented using efficient techniques for finding minimum cuts in graphs; this greatly facilitates incorporation of cross-sentence contextual constraints.
Information systems and the Internet have facilitated the creation of used-product markets that feature a dramatically wider selection, lower search costs, and lower prices than their brick-and-mortar counterparts do. The increased viability of these used-product markets has caused concern among content creators and distributors, notably the Association of American Publishers and Author’s Guild, who believe that used-product markets will significantly cannibalize new product sales.
This proposition, while theoretically possible, is based on speculation as opposed to empirical evidence. In this paper, we empirically analyze the degree to which used products cannibalize new-product sales for books—one of the most prominent used-product categories sold online. To do this, we use a unique data set collected from Amazon.com’s new and used book marketplaces to measure the degree to which used products cannibalize new-product sales. We then use these estimates to measure the resulting first-order changes in publisher welfare and consumer surplus.
Our analysis suggests that used books are poor substitutes for new books for most of Amazon’s customers. The cross-price elasticity of new-book demand with respect to used-book prices is only 0.088. As a result, only 16% of used-book sales at Amazon cannibalize new-book purchases. The remaining 84% of used-book sales apparently would not have occurred at Amazon’s new-book prices. Further, our estimates suggest that this increase in book readership from Amazon’s used-book marketplace increases consumer surplus by approximately $67.21 million annually. This increase in consumer surplus, together with an estimated $45.05 million loss in publisher welfare and a $65.76 million increase in Amazon’s profits, leads to an increase in total welfare to society of approximately $87.92 million annually from the introduction of used-book markets at Amazon.com.
The creation of online consumer communities to provide product reviews and advice has been touted as an important, albeit somewhat expensive component of Internet retail strategies. In this paper, we characterize reviewer behavior at two popular Internet sites and examine the effect of consumer reviews on firms' sales. We use publicly available data from the two leading online booksellers, Amazon.com and BarnesandNoble.com, to construct measures of each firm's sales of individual books. We also gather extensive consumer review data at the two sites. First, we characterize the reviewer behavior on the two sites such as the distribution of the number of ratings and the valence and length of ratings, as well as ratings across different subject categories. Second, we measure the effect of individual reviews on the relative shares of books across the two sites. We argue that our methodology of comparing the sales and reviews of a given book across Internet retailers allows us to improve on the existing literature by better capturing a causal relationship between word of mouth (reviews) and sales since we are able to difference out factors that affect the sales and word of mouth of both retailers, such as the book's quality. We examine the incremental sales effects of having reviews for a particular book versus not having reviews and also the differential sales effects of positive and negative reviews. Our large database of books also allows us to control for other important confounding factors such as differences across the sites in prices and shipping times.
We consider the problem of classifying documents not by topic, but by overall sentiment, e.g., determining whether a review is positive or negative. Using movie reviews as data, we find that standard machine learning techniques definitively outperform human-produced baselines. However, the three machine learning methods we employed (Naive Bayes, maximum entropy classification, and support vector machines) do not perform as well on sentiment classification as on traditional topic-based categorization. We conclude by examining factors that make the sentiment classification problem more challenging.