Customer Review Data Research Articles

The components of perceived quality and their influence on online re-purchase intention

Purpose This study aims to understand holistic consumer perceptions of quality and their effect on re-purchase intentions by measuring the latent characteristics of online Amazon reviews. Design/methodology/approach Data was collected from entries in the Amazon customer review data set, which explicitly mentions quality, economic evaluation and future purchase intention. The analyses included natural language processing, structural topic modeling and econometric analysis. The study used real-time customer reviews to determine the overall perceived quality, the impact of perceived quality on re-purchase intention and the mediating roles of price consciousness and customer satisfaction. Findings Consumers’ perception of overall quality includes product- and service-related dimensions. Perceived quality influences re-purchase intentions through the mediating role of customer satisfaction. While price consciousness impacts the link between perceived quality and customer satisfaction, it does not affect re-purchase intention. Practical implications The managerial implications emphasize multiple dimensions of quality in the online environment and the role of customer satisfaction in consumers’ online re-purchase intentions. The results also illustrate that price effects are insignificant in influencing re-purchase intentions. Thus, while price cuts may encourage initial purchases, quality and customer satisfaction are vital to stimulate re-purchase. Originality/value The e-commerce literature lacks a comprehensive and rigorous understanding of the components of consumers’ perceived quality. This research develops a thorough understanding of what impacts overall e-commerce quality based on real-time customer reviews, avoiding the biases arising from traditional methods, including surveys.

Evaluating Airline Passengers’ Satisfaction during the COVID-19 Pandemic: A Case Study of AirAsia Services through Sentiment Analysis and Topic Modelling

AirAsia has emerged as a dominant force among prominent low-cost airlines in recent years. However, the COVID-19 pandemic outbreak has severely impacted airline services, including AirAsia. There is a strong need for airline services to monitor customer experience and satisfaction from online customer reviews on the website to keep pace with changing customer perceptions toward their service quality. A growing number of travelers choose to express their experiences and emotions on online customer review platforms, resulting in substantial online airline service evaluations. The research analyzes 796 online customer reviews from Skytrax, a well-known online airline review website. The information hidden in customer-generated reviews is analyzed with the text mining technique, including topic modeling and sentiment analysis. The research uses the Latent Dirichlet Allocation (LDA) model for topic analysis and the Valence Aware Dictionary for Sentiment Reasoning (VADER) model for sentiment analysis. The sentiment ratio for AirAsia’s online reviews is approximately 59% positive and 41% negative. Only four reviews are neutral. The findings indicate that the online review of AirAsia has a greater proportion of positive sentiments than negative sentiments. In addition, the topic modeling shows hidden topics with the top high-probability keywords concerned with interior and seat, baggage, online service, staff service, flight schedule, and refund. The research demonstrates using sentiment analysis and topic modeling on customer review data as a more thorough alternative to survey-based models for researching airline service. The research contributes to the methodological advancements in text mining analysis and expands the current knowledge of customer review data.

Determinants of Business Resilience in the Restaurant Industry During the COVID-19 Pandemic: A Textual Analytics Study on an O2O Platform Case

Understanding the resilience capabilities of restaurant operations and the determinants affecting these capabilities is critical to helping restaurants overcome the hardships owing to the coronavirus disease (COVID-19) pandemic. This article adopts a textual analytics approach to scientifically measure consumption trends and identify the shock to restaurant sales using online customer review data from Dianping.com (an O2O platform in China). Moreover, the article proposes a theoretical model of business resilience for the restaurant industry in the context of the pandemic. Then, an empirical investigation on how the determinants in our theoretical framework affect the resilience of restaurant business operations using the panel logit model is conducted. Our findings indicate that the pandemic has severely disrupted the full-service restaurants as compared to the quick-service restaurants. We identify four determinants of resilience, namely social capital (i.e., restaurant rating), physical capital (i.e., contactless service), economic capital (i.e., chain operation), and natural capital (e.g., location), which are significantly associated with the resilience of restaurant business during the pandemic. These four determinants play different roles in the resilience of full-service and quick-service restaurants. The findings of this study have theoretical contribution and generate some important managerial implications for helping the restaurant industry recover from disruptions brought by the COVID-19 pandemic.

Comparison of Ensemble Learning Methods for Mining the Implementation of the 7 Ps Marketing Mix on TripAdvisor Restaurant Customer Review Data

The 7P marketing mix encompasses various business facets, notably the Process element governing internal operations from production to customer service. With the surge in online customer feedback, assessing machine learning efficacy, especially ensemble learning, in classifying 7P-related customer review data has gained prominence. This research aims to fill a gap in existing literature by evaluating ensemble learning’s performance on 7P classification, an area not extensively explored despite prior sentiment analysis studies. Employing a methodology merging Natural Language Processing (NLP) with ensemble learning, the study processes restaurant reviews using NLP techniques and employs ensemble learning for precision and accuracy. Findings demonstrate that DESMI yielded the highest performance metrics with accuracy at 0.697, precision at 0.699, recall at 0.697, and an F1-score of 0.684. These outcomes underscore ensemble learning's potential in handling complex datasets, signifying its relevance for marketers and researchers seeking comprehensive insights from customer reviews within the 7P marketing mix domain. This study sheds light on how ensemble learning outperforms its foundational methods, indicating its prowess in extracting meaningful insights from diverse and intricate customer feedback.

Implementation of Dynamic Topic Modeling to Discover Topic Evolution on Customer Reviews

Annotation and analysis of online customer reviews were identified as significant problems in various domains, including business intelligence, marketing, and e-governance. In the last decade, various approaches based on topic modeling have been developed to solve this problem. The known solutions, however, often only work well on content with static topics. As a result, it is challenging to analyze customer reviews that include dynamic and constantly expanding collections of short and noisy texts. A method was proposed to handle such dynamic content. The proposed system applied a dynamic topic model using BERTopic to monitor topics and word evolution over time. It would help decide when the topic model needs to be retrained to capture emerging topics. Several experiments were conducted to test the practicality and effectiveness of the proposed framework. It demonstrated how a dynamic topic model could handle the emergence of new and over-time-correlated topics in customer review data. As a result, improved performance was achieved compared to the baseline static topic model, with 25% of new segmented texts discovered using the dynamic topic model. Experimental results have, therefore, convincingly demonstrated that the proposed framework can be used in practice to develop automatic review annotation tools.

Positive Relationships between Service Performance and Social Media Use in Internet Retailing

The purpose of this research is to investigate whether differences in Internet retailer service performance influence the use of social media. Information economic theory is used to hypothesize a positive relationship between performance and social media use. To test the relationship, 516 Internet retailers’ websites were investigated, and their service performance was analyzed based on online customer reviews. Exploratory factor analysis transformed the customer review data into refined measurements. The logistic regression model validated the positive relationship between service performance and social media use. Internet retailers who use social media show higher service performance. However, the significant effect is limited to online service performance. Offline service performance (i.e., order fulfillment service performance) was not significantly associated with social media use. This study considers information symmetry as a critical factor in social media use, which differs from previous studies based on traditional technology acceptance frameworks.

Analysis of pet‐food customer postpurchase experience using online customer reviews: Implications for product and marketing strategies

AbstractPet‐food industry growth and demand for specialized product offerings have generated new opportunities for companies to enhance their competitiveness and profitability through effective product differentiation. Recent rise in e‐commerce and technological advancements for capturing and analyzing online customer review data provide new opportunities for large‐scale assessment of customer perceptions of product attributes. This paper presents an integrated qualitative and quantitative approach for utilizing online customer review data to generate insights for informing pet‐food industry product and marketing decisions. The results from the analysis of 5632 online reviews of two pet‐food products highlight (i) product attributes that effectively differentiate (or put on par) two competing pet‐food products, (ii) identify product attributes that are most important for customer postpurchase experience, and (iii) estimate the effect of specific attributes on customers' positive (or negative) postpurchase experience. Specific pet‐food attributes examined include packaging, health/benefit, ingredient, taste, smell, appearance/form, natural/organic, processing, sourcing, small/large breed, price, and service. The advantages of this approach include its ability to leverage large amounts of easily accessible data and its generalizability to the analysis of other consumer goods that have significant online sales and customer review data [EconLit Citations: Q13, M31].

Research on the weak demand signal identification model of innovative product based on domain ontology construction

PurposeThis paper investigates the semantic association mechanisms of weak demand signals that facilitate innovative product development in terms of conceptual and temporal precedence, despite their inherent ambiguity and uncertainty.Design/methodology/approachTo address this challenge, a domain ontology approach is proposed to construct a customer demand scenario-based framework that eliminates the blind spots in weak demand signal identification. The framework provides a basis for identifying such signals and introduces evaluation indices, such as depth, novelty and association, which are integrated to propose a three-dimensional weak signal recognition model based on domain ontology that outperforms existing research.FindingsEmpirical analysis is carried out based on customer comments of new energy vehicles on car platform such as “Auto Home” and “Bitauto”. Results demonstrate that in terms of recognition quantity, the three-dimensional weak demand signal recognition model, based on domain ontology, can accurately identify six demand weak signals. Conversely, the keyword analysis method exhibits a recognition quantity of four weak signals; in terms of recognition quality, the three-dimensional weak demand signal recognition model based on domain ontology can exclude non-demand signals such as “charging technology”, while keyword analysis methods cannot. Overall, the model proposed in this paper has higher sensitivity.Originality/valueThis paper proposes a novel method for identifying weak demand signals that considers the frequency of the signal's novelty, depth and relevance to the target demand. To verify its effectiveness, customer review data for new energy vehicles is used. The results provide a theoretical reference for formulating government policies and identifying weak demand signals for businesses.

Co-creating affective atmospheres in retail experience

Despite substantial research focusing on retail atmospheric elements and the creation of social atmospheres, a holistic theory of atmosphere addressing consumer heterogeneity remains elusive. This research examines how embodied interactions with retail socio-material elements and consumers’ expertise levels create affective atmospheres. Through a three-year-long ethnographic study of winery tours, we identify three core elements in the co-creation of affective atmospheres: material features, sensory modalities, and social interactions. We use text mining to build corresponding dictionaries and analyze customer review data from several retail sites, demonstrating the positive effect of retail atmospherics on customer sentiment. Focusing on consumer segmentation, we further show that different levels of consumer expertise (novices, enthusiasts, and experts) influence the subjective perception of atmospheric elements and the way staff members are expected to behave to jointly create affective atmospheres. Based on our multi-method findings, we offer theoretical and managerial implications that are generalizable to broader retailing contexts.

Enhanced review facilitation service for C2C support: machine learning approaches

Purpose The purpose of this study is to present a methodology for enhancing the quality and usefulness of online reviews for prospective customers to investigate how this contemporary form of instrumental support can be facilitated to strengthen customer-to-customer support. Design/methodology/approach This study develops an analytics framework with applications of machine learning models using customer review data from Amazon.com. Linear regression is commonly used for review helpfulness and sales prediction. In this study, Random Forest model is applied because of its strong performance and reliability. To advance the methodology, a custom script in Python is created to generate Partial Dependence Plots for intensive exploration of the dependency interpretations of review helpfulness and sales. The authors also apply K-Means to cluster reviewers and use the results to generate reviewer qualification scores and collective reviewer scores, which are incorporated into the review facilitation process. Findings The authors find the average helpfulness ratio of the reviewer as the most important determinant of reviewer qualification. The collective reviewer qualification for a product created based on reviewers’ characteristics is found important to customers’ purchase intentions and can be used as a metric for product comparison. Practical implications The findings of this study suggest that service improvement efforts can be performed by developing software applications to monitor reviewer qualifications dynamically, bestowing a badge to top quality reviewers, redesigning review sorting interfaces and displaying the consumer rating distribution on the product page, resulting in improved information reliability and consumer trust. Originality/value This study adds to the research on customer-to-customer support in the service literature. As customer reviews perform as a contemporary form of instrumental support, the authors validate the determinants of review helpfulness and perform an intensive exploration of its dependency interpretation. Reviewer qualification and the collective reviewer qualification scores are generated as new predictors and incorporated into the helpfulness-based review facilitation services.

The effects of customer online reviews on sales performance: The role of mobile phone’s quality characteristics

Quality characteristics have been increasingly emphasized by companies to attract more customers. Previous literature has pointed out the importance of measuring customer’s attitude towards quality characteristics. In addition, the quality characteristics’ classification needs to be improved. Also, few studies have examined the impact of products’ different quality characteristics on sales rank. This paper measures customers’ satisfaction in online reviews based on a sentimental lexicon. The mobile phones’ quality characteristics are classified into five categories based on Kano model. The impact of these categories on sales performance is shown by using Generalized Method of Moments (GMM) model. Our sample is four mobile phone brands’ sales information and customer review data on Amazon websites. Our study extracts quality characteristics and transforms each review into a structured vector. LSTM’s weights are adopted to map quality characteristics into Kano model’s categories. The regression results show that the Excitement and Performance quality characteristics can significantly enhance product’s popularity. We also find that the Indifferent and Reverse quality characteristics can’t have a significant influence on these products. This study also highlights that the Must quality characteristics are essential but can’t continuously improve product’s sales performance. Our findings suggest that products’ quality characteristics may have a disparate influence on purchase decision. It also enlightens managers to pay more attention to important quality characteristics.

Analisis Sentimen Customer Feedback Tokopedia Menggunakan Algoritma Naïve Bayes

Products and customers have a close relationship, therefore UMKM need to build good relationships with customers. The most common way that companies or UMKM do is to look at the reviews given, this is called customer feedback. The results of customer feedback to companies or UMKM can improve service and product quality. The problem that arises is how to process the many reviews given, especially reviews from marketplaces like Tokopedia. Therefore, a method is needed to see user reviews of the products being sold, whether positive or negative. The method that will be used is sentiment analysis. Sentiment analysis is the process of understanding and extracting and automatically processing text data and can produce sentiments that are displayed in a sentence. The steps taken were taking House of Smith customer review data at Tokopedia, manual labeling to get positive and negative data reviews, data preprocessing, TF-IDF weighting and classification using the Naïve Bayes algorithm. The results of sentiment testing using the Naïve Bayes algorithm with TF-IDF weighting quality accuracy of 83% with visualization of the distribution of words that appear the most are the words 'good', 'comfortable' and 'use' for positive reviews. The most frequent negative reviews were 'material' and 'thin' which indicated that some buyers felt that the product had a thin material.

Customer sentiment analysis and prediction of halal restaurants using machine learning approaches

PurposeThere is a strong prerequisite for organizations to analyze customer review behavior to evaluate the competitive business environment. The purpose of this study is to analyze and predict customer reviews of halal restaurants using machine learning (ML) approaches.Design/methodology/approachThe authors collected customer review data from the Yelp website. The authors filtered the reviews of only halal restaurants from the original data set. Following cleaning, the filtered review texts were classified as positive, neutral or negative sentiments, and those sentiments were scored using the AFINN and VADER sentiment algorithms. Also, the current study applies four machine learning methods to classify each review toward halal restaurants into its sentiment class.FindingsThe experiment showed that most of the customer reviews toward halal restaurants were positive. The authors also discovered that all of the methods (decision tree, linear support vector machine, logistic regression and random forest classifier) can correctly classify the review text into sentiment class, but logistic regression outperforms the others in terms of accuracy.Practical implicationsThe results facilitate halal restaurateurs in identifying customer review behavior.Social implicationsSentiment and emotions, according to appraisal theory, form the basis for all interactions, facilitating cognitive functions and supporting prospective customers in making sense of experiences. Emotion theory also describes human affective states that determine motives and actions. The study looks at how potential customers might react to a halal restaurant’s consensus on social media based on reviewers’ opinions of halal restaurants because emotions can be conveyed through reviews.Originality/valueThis study applies machine learning approaches to analyze and predict customer sentiment based on the review texts toward halal restaurants.

Fake Review Detection Using Machine Learning

E-commerce and online shopping is a booming industry, especially during the COVID-19 pandemic. Online shopping allows customers to browse and purchase products from home using just a phone or laptop. Customers often never end up seeing their product in real life before the purchase, and are instead dependent on photos and descriptions uploaded by the seller. Thus comes the need for customer reviews: an evaluation of a product by other buyers which can influence other shoppers’ decisions to make the purchase. However, reviews can be tainted to provide a fake or unrealistic depiction of a product. Sellers can pay people or robots to leave fake reviews under competitors or their own stores to increase/decrease sales turnout. Such reviews can be harmful to the buyer or other sellers, often ending up with an unhappy customer. Using supervised datasets consisting of real and fake reviews, we can train a variety of machine learning and deep learning models to recognize attributes differentiating between the two types of reviews. Big e-commerce platforms such as Amazon, Yelp, and Tripadvisor are all common targets of fake reviews, and the implementation of fake review detection could create a more assuring shopping experience for customers. In this paper, we analyze and break down customer review data and attempt to build models that form conclusions from it.

Sentiment Anlysis On Customer Reviews Using Support Vector Machine and Usability Scoring Using System Usability Scale

Al-Ghiff Steak is a restaurant located in Cirebon City that offers quality steaks at affordable prices. For maintaining a competitive Al-Ghiff Steak advantage and reputation, it is important to build a good relationship with customers and have a business strategy that considers customer opinions. However, in its implementation, Al-Ghiff Steak has difficulty when collecting and processing customer review data manually. Therefore, it is necessary to conduct sentiment analysis by utilizing Google Reviews to determine customer perspectives regarding Al-Ghiff Steak products and services. This analysis was conducted on 968 Google Review reviews from 2016 to 2020 using the Support Vector Machine (SVM) and Term Frequency-Inverse Document Frequency (TF-IDF) methods. Classification testing is done with a confusion matrix against four parameters: accuracy, precision, recall, and f1-score. SVM with TF-IDF gets accuracy value 83%, precision 64%, recall 60% and f1-score 59%. The sentiment classification result is then visualized in the form of a dashboard. We utilize the System Usability Scale (SUS) for usability testing, which produces a value of 77.5. This result achieve the Acceptable category and an Excellent rating.

Sentimen Analisis Customer Review Produk Shopee Indonesia Menggunakan Algortima Naïve Bayes Classifier

Gaining customer satisfaction and trust has become the main challenge in achieving success in the business world. Business people need to identify problems that arise from reviews given by customers. However, reading and classifying each review takes a long time and is considered ineffective. To overcome this, this study aims to analyze the customer sentiment of shopee products using the nave Bayes classifier algorithm. The data used in this study is a customer review of the Xiaomi Redmi Note 9 products which are sold on the Shopee Indonesia website. Customer review data is collected by applying the Web Scraping technique. The algorithm used in this study is the Naïve Bayes Classifier which is known to be popular and effective in classifying data. This study also applies the Knowledge Discovery in Text (KDT) methodology to extract information from text data. The results of the classification using the Naïve Bayes algorithm found an accuracy value of 85%. This study proves that by applying sentiment analysis techniques, business people are able to find out the opinions of customers as an evaluation material that needs to be done to optimize the products and services provided.

Social Media-Enabled Customer Co-Creation Experience Value: Conceptualisation and Measurement

ABSTRACT Despite the dominance of customer co-creation experience value (CCEV) in the value co-creation research, its theoretical domain has not been systematically analysed and developed. Specifically, little is known about the nature of social media-enabled CCEV. To address the knowledge gaps, this study employs a five-step procedure to develop a conceptualisation and instrument of CCEV construct in a social media context. First, drawing on the relevant literature, we determine the conceptual framework, definition, and assumptions of CCEV. Second, we employ grounded theory methodology to analyse customer review data from three representative social media sites and identify the dimensions of CCEV (four themes consisting of 10 dimensions). Third, we create an item pool of CCEV based on the prior work and in-depth interviews of 10 customers. In the fourth and fifth steps, we collect three waves of quantitative data to develop and validate the instrument including item refinement (n1 = 234), instrument validation (n2 = 521), and nomological validity examination (n3 = 615). The results indicate that the instrument exhibits satisfactory reliability and validity. Our work contributes to theory and practice by presenting a context-specific and fine-grained conceptualisation of CCEV and a reliable and valid survey instrument.

Do Hedonic or Utilitarian Types of Online Product Reviews Make Reviews More Helpful?

Online sales can be influenced significantly by customer reviews, and thus there are several studies on what makes an online review helpful to consumers. However, none of those researches address the review helpfulness in the context of hedonic and utilitarian review types. This study examines how product type (hedonic and utilitarian) moderates the relationship between the level of review type (hedonicity and utilitarianity) and review helpfulness. To test the moderating effects, customer review data for perfume and bar soap product was collected from Amazon.com and analyzed by using a text-mining tool (QDA Miner) and a structural equation modeling software (AMOS 22.0). The results of this study indicate that the product type moderates the impact of the review type on review helpfulness when product type and review type are incongruent. Further, results show that the number of sentences in a customer review affects the review helpfulness when product type is utilitarian.

Do Hedonic or Utilitarian Types of Online Product Reviews Make Reviews More Helpful?

Online sales can be influenced significantly by customer reviews, and thus there are several studies on what makes an online review helpful to consumers. However, none of those researches address the review helpfulness in the context of hedonic and utilitarian review types. This study examines how product type (hedonic and utilitarian) moderates the relationship between the level of review type (hedonicity and utilitarianity) and review helpfulness. To test the moderating effects, customer review data for perfume and bar soap product was collected from Amazon.com and analyzed by using a text-mining tool (QDA Miner) and a structural equation modeling software (AMOS 22.0). The results of this study indicate that the product type moderates the impact of the review type on review helpfulness when product type and review type are incongruent. Further, results show that the number of sentences in a customer review affects the review helpfulness when product type is utilitarian.

Analisis Sentimen E-Wallet di Twitter Menggunakan Support Vector Machine dan Recursive Feature Elimination

Grouping of positive or negative sentiments in text reviews is increasingly being done automatically for identification. The selection of features in the classification is a problem that is often not solved. Most of the feature selection related to sentiment classification techniques is insurmountable in terms of evaluating significant features that reduce classification performance. Good feature selection technique can improve sentiment classification performance in machine learning approach. First, two sets of customer review data are labeled with sentiment and then retrieved, processed for evaluation. Next, the supports vector machine (svm-rfe) method is created and tested on the dataset. Svm-rfe will be run to measure the importance of the feature by rating the feature iteratively. For sentiment classification, only the top features of the ranking feature sequence will be used. Finally, performance is measured using accuracy, precision, recall, and f1-score. The experimental results show promising performance with an accuracy rate of 81%. This level of reduction is significant in making optimal use of computing resources while maintaining the efficiency of classification performance