Review Dataset Research Articles

GraphIC: A graph-based approach for identifying complaints from code-mixed product reviews

Enterprises and online merchants can benefit from automatically detecting complaints from social media platforms and shopping websites. Prior studies on automatically identifying complaints on social media have relied chiefly on analyzing complaints in English, with little consideration for code-mixed complaints. We manually annotated the emotion and sentiment classes in the Product Review dataset, a code-mixed language complaint corpus. The extended corpus consists of 3711 instances annotated with the complaint, sentiment, and emotion labels. We present a Graph Attention Network (GAT) based multi-task framework that intends to simultaneously learn three closely related tasks, complaint detection (primary task), sentiment classification, and emotion recognition (auxiliary task). Based on a quantitative and qualitative study on the extended version of the Product Review corpus, the proposed approach attains an accuracy of 72.82% and a macro-F1 of 71% for the intended task of recognizing complaints, outperforming several single-task, multi-task baselines.11Resources are available at: https://github.com/appy1608/ESWA2022_GraphIC.

A novel hybrid deep learning model for aspect based sentiment analysis

SummaryThe usage of social media, forums, and e‐commerce websites have been widely increased. Feedback from customers has a big impact on the final product. A service provider, merchant, or manufacturer need all the information, even if it is just a comment or a review about a service or a product. So, it is vital to look at input from users, and therefore sentiment analysis has received a lot of interest. Sentiment analysis is a method for identifying and analyzing text in order to determine the features, qualities, and viewpoints of particular user. Extracting user aspects is the main part of this process, and it is used to group the user aspects. In recent years, convolutional neural network (CNN) models have gained popularity in natural language processing. Thus, this research proposes a novel hybrid CNN model by concatenating the bidirectional long short‐term memory and CNN models to process the data sequentially by learning their high‐level features. The concatenated method minimizes the loss of critical information. Benchmark product reviews and hotel review datasets are employed in the experiments, and accuracies of 93.6% for the product review dataset and 92.7% for the hotel review dataset are achieved by the proposed hybrid model when compared to state‐of‐the‐art techniques.

A Book-Influence-Evaluation Method Based on User Ratings of E-Commerce Platform

In online social networks, finding high-influence nodes is a crucial component of complex network research. A new book impact evaluation method based on user rating is proposed in this research for the social network created by the buying and selling behaviors on the e-commerce platform. It intends to rank the book nodes in accordance with customer feedback data following user purchases. The method calculates the influence score of a book by predicting its popularity based on user evaluations of the book. To verify the validity and accuracy of the method, the research analyzes a real review dataset from Amazon, a large e-commerce platform, and designs two comparison experiments with different time spans and compares them with five other web analytics metrics. The experimental findings show that the method is efficient and precise in evaluating the influence of book nodes.

A fine-grained deep learning model using embedded-CNN with BiLSTM for exploiting product sentiments

As technology advances, Facebook, Twitter, and microblogging sites have become the most effective platforms for communication and information exchange. Through these forums, people can share their views and experiences. These platforms enable discussion about a certain product that can be a valuable resource used to inform any decision-making process. For such studies, the majority of advanced-level researchers employed deep learning and machine learning models in conjunction with natural language processing (NLP). In recent years, the use of pre-trained models, such as Glove and BERT, in aspect-based sentiment analysis (ABSA) has increased. In ABSA, the auxiliary information is required to distinguish each aspect of this fine-grained task. However, BERT’s input format is restricted to a collection of words that cannot include more context knowledge. To address this problem, a BiLSTM and embedded CNN-based deep learning model has been presented for sentiment analysis at the aspect level. Initially, datasets were compiled from several sources. Then, an auxiliary feature was extracted using standard NLP. The auxiliary features were further refined and transformed into feature vectors based on the proposed embedded CNN model. Finally, a BiLSTM-based classification has been performed for sentiment classification. The experimental evaluation demonstrated that the performance of the suggested technique achieves on the SemEval dataset in terms of F1 score and accuracy by 81.7 and 83.3 percentage points, respectively, and on the product review dataset by 80.8 and 83.1 percentage points, respectively.

Exploring success factors of tourism performing arts by analyses of online reviews

在线点评分析探析旅游演艺成功因素研究目的本研究旨在通过分析大型在线评论数据集来探索旅游表演艺术项目的成功因素。研究设计/方法/途径共收集和预处理来自携程网的 195,230 条评论。利用深度学习方法来估计单词之间的相似性。然后, 进行回归分析以确定成功因素。研究结果本研究提取了影响游客对旅游表演艺术满意度的四个积极因素和两个消极因素。结果表明, 游客最关注中国传统文化方面、视听效果和演员的表演热情。研究限制/影响尽管我们的数据集很大, 但它只关注中文评论。比较不同国家提供的旅游表演艺术项目的成功因素将是有用和有趣的未来研究方向。研究实践意义研究结果有助于发展旅游表演艺术项目, 以吸引游客前往旅游目的地。研究原创性/价值研究表明, 通过文本挖掘技术分析旅游表演艺术的在线评论是了解游客满意度的有效方法。

Been There, Done That: How Episodic and Semantic Memory Affects the Language of Authentic and Fictitious Reviews

AbstractThis article suggests a theory-driven approach to address the managerial problem of distinguishing between real and fake reviews. Building on memory research and linguistics, we predict that when recollecting an authentic experience in a product review, people rely to a greater extent on episodic memory. By contrast, when writing a fictitious review, people do not have episodic memory available to them. Therefore, they must rely to a greater extent on semantic memory. We suggest that reliance on these different memory types is reflected in the language used in authentic and fictitious reviews. We develop predictions about five linguistic features characterizing authentic versus fictitious reviews. We test our predictions via a multi-method approach, combining computational linguistics, experimental design, and machine learning. We employ a large-scale experiment to derive a dataset of reviews, as well as two datasets containing reviews from online platforms. We also test whether an algorithm relying on our theory-driven linguistic features is context independent, relative to other benchmark algorithms, and shows better cross-domain performance when tested across datasets. By developing a theory that extends memory and psycholinguistics research to the realm of word of mouth, this work contributes to our understanding of how authentic and fictitious reviews are created.

Arabic aspect sentiment polarity classification using BERT

Aspect-based sentiment analysis (ABSA) is a textual analysis methodology that defines the polarity of opinions on certain aspects related to specific targets. The majority of research on ABSA is in English, with a small amount of work available in Arabic. Most previous Arabic research has relied on deep learning models that depend primarily on context-independent word embeddings (e.g. word2vec), where each word has a fixed representation independent of its context. This article explores the modeling capabilities of contextual embeddings from pre-trained language models, such as BERT, and making use of sentence pair input on Arabic aspect sentiment polarity classification task. In particular, we develop a simple but effective BERT-based neural baseline to handle this task. Our BERT architecture with a simple linear classification layer surpassed the state-of-the-art works, according to the experimental results on three different Arabic datasets. Achieving an accuracy of 89.51% on the Arabic hotel reviews dataset, 73.23% on the Human annotated book reviews dataset, and 85.73% on the Arabic news dataset.

Hierarchical Interpretation of Neural Text Classification

Abstract Recent years have witnessed increasing interest in developing interpretable models in Natural Language Processing (NLP). Most existing models aim at identifying input features such as words or phrases important for model predictions. Neural models developed in NLP, however, often compose word semantics in a hierarchical manner. As such, interpretation by words or phrases only cannot faithfully explain model decisions in text classification. This article proposes a novel Hierarchical Interpretable Neural Text classifier, called HINT, which can automatically generate explanations of model predictions in the form of label-associated topics in a hierarchical manner. Model interpretation is no longer at the word level, but built on topics as the basic semantic unit. Experimental results on both review datasets and news datasets show that our proposed approach achieves text classification results on par with existing state-of-the-art text classifiers, and generates interpretations more faithful to model predictions and better understood by humans than other interpretable neural text classifiers.1

Global reconstruction of language models with linguistic rules – Explainable AI for online consumer reviews

Analyzing textual data by means of AI models has been recognized as highly relevant in information systems research and practice, since a vast amount of data on eCommerce platforms, review portals or social media is given in textual form. Here, language models such as BERT, which are deep learning AI models, constitute a breakthrough and achieve leading-edge results in many applications of text analytics such as sentiment analysis in online consumer reviews. However, these language models are “black boxes”: It is unclear how they arrive at their predictions. Yet, applications of language models, for instance, in eCommerce require checks and justifications by means of global reconstruction of their predictions, since the decisions based thereon can have large impacts or are even mandatory due to regulations such as the GDPR. To this end, we propose a novel XAI approach for global reconstructions of language model predictions for token-level classifications (e.g., aspect term detection) by means of linguistic rules based on NLP building blocks (e.g., part-of-speech). The approach is analyzed on different datasets of online consumer reviews and NLP tasks. Since our approach allows for different setups, we further are the first to analyze the trade-off between comprehensibility and fidelity of global reconstructions of language model predictions. With respect to this trade-off, we find that our approach indeed allows for balanced setups for global reconstructions of BERT’s predictions. Thus, our approach paves the way for a thorough understanding of language model predictions in text analytics. In practice, our approach can assist businesses in their decision-making and supports compliance with regulatory requirements.

Likelihood estimation of sparse topic distributions in topic models and its applications to Wasserstein document distance calculations

This paper studies the estimation of high-dimensional, discrete, possibly sparse, mixture models in the context of topic models. The data consists of observed multinomial counts of p words across n independent documents. In topic models, the p×n expected word frequency matrix is assumed to be factorized as a p×K word-topic matrix A and a K×n topic-document matrix T. Since columns of both matrices represent conditional probabilities belonging to probability simplices, columns of A are viewed as p-dimensional mixture components that are common to all documents while columns of T are viewed as the K-dimensional mixture weights that are document specific and are allowed to be sparse. The main interest is to provide sharp, finite sample, ℓ1-norm convergence rates for estimators of the mixture weights T when A is either known or unknown. For known A, we suggest MLE estimation of T. Our nonstandard analysis of the MLE not only establishes its ℓ1 convergence rate, but also reveals a remarkable property: the MLE, with no extra regularization, can be exactly sparse and contain the true zero pattern of T. We further show that the MLE is both minimax optimal and adaptive to the unknown sparsity in a large class of sparse topic distributions. When A is unknown, we estimate T by optimizing the likelihood function corresponding to a plug in, generic, estimator Aˆ of A. For any estimator Aˆ that satisfies carefully detailed conditions for proximity to A, we show that the resulting estimator of T retains the properties established for the MLE. Our theoretical results allow the ambient dimensions K and p to grow with the sample sizes. Our main application is to the estimation of 1-Wasserstein distances between document generating distributions. We propose, estimate and analyze new 1-Wasserstein distances between alternative probabilistic document representations, at the word and topic level, respectively. We derive finite sample bounds on the estimated proposed 1-Wasserstein distances. For word level document-distances, we provide contrast with existing rates on the 1-Wasserstein distance between standard empirical frequency estimates. The effectiveness of the proposed 1-Wasserstein distances is illustrated by an analysis of an IMDB movie reviews data set. Finally, our theoretical results are supported by extensive simulation studies.

Using neutral sentiment reviews to improve customer requirement identification and product design strategies

A clear understanding of customer needs is key to the success of product design strategies. Traditional methods of understanding customer needs rely on costly marketing surveys and have difficulties accurately capturing new customer requirements in a fast-evolving market. Online reviews with positive and negative sentiments are commonly used as effective sources for mining customer requirements. Previous studies related to product design strategies often overlook neutral sentiment when analyzing online reviews for product design improvement. In this study, we propose a customer requirement identification framework that identifies the product attributes reflecting customer needs from online reviews, considering three types of sentiment polarities: positive, negative, and neutral sentiment. We categorize the identified customer needs into five product attribute categories that help form product design strategies using the Kano model. Evaluations using two review datasets for laptops and smartphones show that the consideration of neutral reviews caused the majority of the product attributes to be categorized differently by the proposed method. Furthermore, the product categorization obtained from our method achieved a better agreement with domain experts and consumers than that produced by the baseline Kano model, not considering neutral sentiment. We further established evidence that the product design informed by our product categorization results achieved better customer satisfaction than those generated from the baseline Kano model and the questionnaire-based Kano model.

Visual Enhancement Capsule Network for Aspect-based Multimodal Sentiment Analysis

Multimodal sentiment analysis, which aims to recognize the emotions expressed in multimodal data, has attracted extensive attention in both academia and industry. However, most of the current studies on user-generated reviews classify the overall sentiments of reviews and hardly consider the aspects of user expression. In addition, user-generated reviews on social media are usually dominated by short texts expressing opinions, sometimes attached with images to complement or enhance the emotion. Based on this observation, we propose a visual enhancement capsule network (VECapsNet) based on multimodal fusion for the task of aspect-based sentiment analysis. Firstly, an adaptive mask memory capsule network is designed to extract the local clustering information from opinion text. Then, an aspect-guided visual attention mechanism is constructed to obtain the image information related to the aspect phrases. Finally, a multimodal fusion module based on interactive learning is presented for multimodal sentiment classification, which takes the aspect phrases as the query vectors to continuously capture the multimodal features correlated to the affective entities in multi-round iterative learning. Otherwise, due to the limited number of multimodal aspect-based sentiment review datasets at present, we build a large-scale multimodal aspect-based sentiment dataset of Chinese restaurant reviews, called MTCom. The extensive experiments both on the single-modal and multimodal datasets demonstrate that our model can better capture the local aspect-based sentiment features and is more applicable for general multimodal user reviews than existing methods. The experimental results verify the effectiveness of our proposed VECapsNet.

Potential Association of Isolated γ-Glutamyltransferase Elevation with Incident Ischemic Heart Disease in Lean Koreans.

Isolated elevation of γ-glutamyltransferase (GGT), a microsomal membrane-bound protein, is commonly observed in non-obese Koreans without diabetes, and its clinical implications are not well-known. Therefore, we aimed to investigate the longitudinal effect of isolated GGT on the incidence of ischemic heart disease (IHD) risk in a large cohort of lean non-diabetic Koreans. Data were obtained from the Health Risk Assessment Study (HERAS) and Korea Health Insurance Review and Assessment (HIRA) datasets. The participants were divided into four groups according to the GGT quartile after the exclusion of those participants with diabetes, a body mass index (BMI) ≥ 25 kg/m2, alanine aminotransferase (ALT) ≥ 40 IU/L, and aspartate aminotransferase (AST)/ALT > 1.5, as well as those positive for hepatitis B surface antigen or hepatitis C antibody. We prospectively assessed the hazard ratios (HRs) with 95% confidence intervals (CIs) for IHD using multivariate Cox proportional hazard regression models over a 50-month period. During the follow-up period, 183 individuals (1.85%) developed IHD. After setting the lowest GGT quartile as a reference group, the HRs of IHD for GGT quartiles 2-4 were 1.66 (95% CI 0.95-2.89), 1.82 (95% CI 1.05-3.16), and 1.98 (95% CI 1.12-3.50), respectively, after adjusting for age, sex, body mass index, smoking status, alcohol consumption, physical activity, mean arterial blood pressure, fasting plasma glucose, and dyslipidemia. An isolated high GGT may be an additional measure for assessing and managing future IHD risks among lean Koreans without diabetes.

Opinion Mining Using Normal Discriminant Piecewise Regressive (NDPR) Sentiment Classification Technique

Sentiment analysis through opinion mining is determined through significant and growing interest for many industries including hotel, tourism, educations and so on. Sentiment analysis includes design of the system to search the user opinions in blog posts, comments, reviews or tweets regarding the product, policy or area. Many researchers carried out their research on opinion mining to identify the polarity of the statements. But the main problem during opinion mining is that the words chosen do not solve attribute relevancy and could not classify the positive and negative usage of uncertain terms. In order to address these problems, normal discriminant piecewise regressive (NDPR) sentiment classification technique is introduced. NDPR technique perform three processes, namely, pre-processing, feature extraction and classification to improve the accuracy level through forming classes (i.e., positive, neutral and negative) based on the extracted words from user review comments. Initially, NDPR technique performs the data pre-processing task for stemming and removing the stop words from review statements to reduce the file size that in turn improves the efficiency. After that, normal discriminant feature extraction process is carried out in NDPR technique to extract the opinion word from the review statements sent by reviewers. The related opinion words are systematized for their semantic equivalence of sentiment based on extracted word. This helps to reduce the time consumption to extract the opinions from reviewers. Finally, piecewise regressive sentiment classification (PRSC) process is carried out in NDPR technique to analyse the semantic opinion words for evaluating the sentiment class label. The sentiment class labels are categorized into positive, neutral and negative sentiments with the user review comments. This in turn helps to reduce the time consumption to extract the opinions from reviewers and to improve the review detection accuracy. The performance evaluation of NDPR technique is carried out with standard benchmark datasets of consumer product and services reviews extracted. The parameters used in evaluation are number of customer review words, accuracy, time complexity (TC) and false positive rate. Experimental analysis shows that NDPR technique reduces the time to extract the opinions from reviewers and false positive rate.

Computational Intelligence Based Recurrent Neural Network for Identification Deceptive Review in the E-Commerce Domain.

Most consumers depend on online reviews posted on e-commerce websites when determining whether or not to buy a service or a product. Moreover, due to the presence of fraudulent (deceptive) reviews, the fundamental problem in such reviews is not fully addressed. Thus, deceptive reviews present wrong and misguiding opinions that are harmful to consumers and e-commerce. People called fraudsters who intentionally write deceptive reviews to target and deceive potential consumers, as they target businesses that have a well-built reputation or fame for their personal promotion, create such reviews. Therefore, developing a deceptive review detection system is essential for identifying and classifying online product reviews as truthful or fake/deceptive reviews. The main objective of this research work is to analyze and identify online deceptive reviews in electronic product reviews in the Amazon and Yelp domains. For this purpose, two experiments were conducted individually. The first was executed on standard Yelp product reviews. The second was performed on Amazon product review datasets. For this dataset, we created and labeled it using a deceptiveness score calculated based on features extracted from the review text using the linguistic inquiry and word count (LIWC) tool. These features were authenticity, negative words, comparing words negation words, analytical thinking, and positive words as well as the given rating value by a user. The recurrent neural network, bidirectional long short-term memory (RNN-BLSTM) model, was used to both datasets in order to conduct the evaluation. The application of this model was contingent upon the learning of words embedding of the review text. Finally, we evaluated the RNN-BLSTM model's performance using the Yelp and Amazon datasets and compared the results. The results were 89.6% regarding testing accuracy for both datasets. From our experimental results, we observed that the LIWC feature with word embedding in the review text provided better accuracy performance compared with other existing methods.

Detecting fake-review buyers using network structure: Direct evidence from Amazon

Online reviews significantly impact consumers' decision-making process and firms' economic outcomes and are widely seen as crucial to the success of online markets. Firms, therefore, have a strong incentive to manipulate ratings using fake reviews. This presents a problem that academic researchers have tried to solve for over two decades and on which platforms expend a large amount of resources. Nevertheless, the prevalence of fake reviews is arguably higher than ever. To combat this, we collect a dataset of reviews for thousands of Amazon products and develop a general and highly accurate method for detecting fake reviews. A unique difference between previous datasets and ours is that we directly observe which sellers buy fake reviews. Thus, while prior research has trained models using laboratory-generated reviews or proxies for fake reviews, we are able to train a model using actual fake reviews. We show that products that buy fake reviews are highly clustered in the product reviewer network. Therefore, features constructed from this network are highly predictive of which products buy fake reviews. We show that our network-based approach is also successful at detecting fake review buyers even without ground truth data, as unsupervised clustering methods can accurately identify fake review buyers by identifying clusters of products that are closely connected in the network. While text or metadata can be manipulated to evade detection, network-based features are more costly to manipulate because these features result directly from the inherent limitations of buying reviews from online review marketplaces, making our detection approach more robust to manipulation.

Model Comparison in Sentiment Analysis: A Case Study of Amazon Product Reviews

Sentiment analysis is essential in NLP, especially in businesses because it can improve customer services. This paper focuses on a particular case of sentiment analysis, a case study of Amazon reviews of books on kindle. Firstly, this paper applies several non-deep-learning algorithms including Logistic Regression, Naïve Bayes, Support Vector Machine, Convolutional Neural Network, and Recurrent Neural Network, and compares their accuracies. Especially, for deep learning methods, this paper studies the slope of accuracies concerning the number of hidden layers. Secondly, as a multi-class text classification problem, the product review data set has five labels ranging from one star to five stars, a new method called Hybrid Sequential Binary Classification (HSBC) is introduced in this paper, which improves the behavior of classical binary classifiers on a multi-class text classification problem. Moreover, a comparison of HSBC and multi-class classification models is presented.

Doc2Vec-based Approach for Extracting Diverse Evaluation Expressions from Online Review Data

This paper proposes a method for extracting diverse expressions from online movie review texts for a given keyword query. When people watch a movie that makes them cry, they generally do not say ``I cried.'' Instead, they use such euphemistic language as ``I needed a handkerchief'' or ``My makeup was running.''To enable information retrieval based on audience reactions such as ``movies that make me cry'' using review texts, various paraphrased expressions must be collected for arbitrary queries. Our proposed method extracts such expressions from review datasets by applying two extensions to Doc2Vec: 1) it changes the granularity of the training sentences to mitigate a lack of context, and 2) it applies query expansion for similarity calculation in advance.We conducted a large-scale crowdsourcing experiment with 1.29 million actual sentences taken from Yahoo! Movies, Japan. The experimental result revealed that changing the training data granularity and adding the query expansion effectively collect more diverse expressions that have a meaning similar to the given query.

Impact of demography on linguistic aspects and readability of reviews and performances of sentiment classifiers

This study investigates how user demographics influence the linguistic characteristics and readability of the reviews and the performance of sentiment classification methods. Two restaurant review datasets representing consumers of distinct demographics, such as English language fluency levels, socio-culture, and geographic locations, are compared. It is observed that demographic dissimilarities affect various linguistic attributes, such as sentence structure and vocabulary usage of reviews. The Mann–Whitney U test indicates significant differences in these two sets of reviews on various linguistic aspects. Regarding the readability of the two corpora, no noticeable difference is perceived by Flesch Reading Ease (FRE) test. The discrepancy in the results of lexicon-based and machine learning-based sentiment classification methods in two corpora suggests that demography has an influential role in the performance of sentiment classifiers.

Optimizing Sentiment Big Data Classification Using Multilayer Perceptron

Internet-based platforms such as social media have a great deal of big data that is available in the shape of text, audio, video, and image. Sentiment Analysis (SA) of this big data has become a field of computational studies. Therefore, SA is necessary in texts in the form of messages or posts to determine whether a sentiment is negative or positive. SA is also crucial for the development of opinion mining systems. SA combines techniques of Natural Language Processing (NLP) with data mining approaches for developing inelegant systems. Therefore, an approach that can classify sentiments into two classes, namely, positive sentiment and negative sentiment is proposed. A Multilayer Perceptron (MLP) classifier has been used in this document classification system. The present research aims to provide an effective approach to improving the accuracy of SA systems. The proposed approach is applied to and tested on two datasets, namely, a Twitter dataset and a movie review dataset; the accuracies achieved reach 85% and 99% respectively.