Customer Review Datasets Research Articles

AEGA: enhanced feature selection based on ANOVA and extended genetic algorithm for online customer review analysis.

Sentiment analysis involves extricating and interpreting people's views, feelings, beliefs, etc., about diverse actualities such as services, goods, and topics. People intend to investigate the users' opinions on the online platform to achieve better performance. Regardless, the high-dimensional feature set in an online review study affects the interpretation of classification. Several studies have implemented different feature selection techniques; however, getting a high accuracy with a very minimal number of features is yet to be accomplished. This paper develops an effective hybrid approach based on an enhanced genetic algorithm (GA) and analysis of variance (ANOVA) to achieve this purpose. To beat the local minima convergence problem, this paper uses a unique two-phase crossover and impressive selection approach, gaining high exploration and fast convergence of the model. The use of ANOVA drastically reduces the feature size to minimize the computational burden of the model. Experiments are performed to estimate the algorithm performance using different conventional classifiers and algorithms like GA, Particle Swarm Optimization (PSO), Recursive Feature Elimination (RFE), Random Forest, ExtraTree, AdaBoost, GradientBoost, and XGBoost. The proposed novel approach gives impressive results using the Amazon Review dataset with an accuracy of 78.60 %, F1 score of 79.38 %, and an average precision of 0.87, and the Restaurant Customer Review dataset with an accuracy of 77.70 %, F1 score of 78.24 %, and average precision of 0.89 as compared to other existing algorithms. The result shows that the proposed model outperforms other algorithms with nearly 45 and 42% fewer features for the Amazon Review and Restaurant Customer Review datasets.

Category-learning attention mechanism for short text filtering

In machine translation, the attention mechanism highlights relevant words according to the distances between the source and target vectors dynamically. However, its ability to optimize text classification is limited because this mechanism only calculates weights inside the same text. There is an uneven distribution of words between ham (not spam) and spam categories, and this category-level feature has not previously been utilized in the attention mechanism for filtering short texts. In addition, short text filtering is uniquely challenging due to the length, sparsity and informal writing of texts, as well as a need for rapid processing. We propose a novel category-level attention mechanism called “category-learning attention,” which highlights words intensely distributed in the same category by dynamically calculating a category differentiation matrix for each short text. The category-learning attention mechanism is extended to the category-learning scaled-dot-product attention and the category-learning multi-head attention (CL-MHA) mechanisms. The CL-MHA mechanism is then applied to a bidirectional gate recurrent unit (Bi-GRU) model for performance evaluation by using the SMS spam collection dataset hosted at the University of California, Irvine. Performance metrics including the accuracy, precision, recall, and F1 score demonstrate that the CL-MHA mechanism significantly improves the performance of Bi-GRU for short text filtering with an accuracy of 99.35%, higher than any previously reported machine learning models. In addition, experiments conducted on three datasets - a Chinese SMS spam dataset, a benchmark movie review dataset, and a benchmark customer review dataset - further validate the effectiveness of the proposed model. The proposed CL-MHA Bi-GRU model has an accuracy of 99.46% when evaluated on the Chinese SMS spam dataset.

Unboxing Deep Learning Model of Food Delivery Service Reviews Using Explainable Artificial Intelligence (XAI) Technique.

The demand for food delivery services (FDSs) during the COVID-19 crisis has been fuelled by consumers who prefer to order meals online and have it delivered to their door than to wait at a restaurant. Since many restaurants moved online and joined FDSs such as Uber Eats, Menulog, and Deliveroo, customer reviews on internet platforms have become a valuable source of information about a company’s performance. FDS organisations strive to collect customer complaints and effectively utilise the information to identify improvements needed to enhance customer satisfaction. However, only a few customer opinions are addressed because of the large amount of customer feedback data and lack of customer service consultants. Organisations can use artificial intelligence (AI) instead of relying on customer service experts and find solutions on their own to save money as opposed to reading each review. Based on the literature, deep learning (DL) methods have shown remarkable results in obtaining better accuracy when working with large datasets in other domains, but lack explainability in their model. Rapid research on explainable AI (XAI) to explain predictions made by opaque models looks promising but remains to be explored in the FDS domain. This study conducted a sentiment analysis by comparing simple and hybrid DL techniques (LSTM, Bi-LSTM, Bi-GRU-LSTM-CNN) in the FDS domain and explained the predictions using SHapley Additive exPlanations (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME). The DL models were trained and tested on the customer review dataset extracted from the ProductReview website. Results showed that the LSTM, Bi-LSTM and Bi-GRU-LSTM-CNN models achieved an accuracy of 96.07%, 95.85% and 96.33%, respectively. The model should exhibit fewer false negatives because FDS organisations aim to identify and address each and every customer complaint. The LSTM model was chosen over the other two DL models, Bi-LSTM and Bi-GRU-LSTM-CNN, due to its lower rate of false negatives. XAI techniques, such as SHAP and LIME, revealed the feature contribution of the words used towards positive and negative sentiments, which were used to validate the model.

Joint Dilated Convolution and Self-attention for Cross-domain Sentiment Analysis

Cross-domain sentiment analysis aims to use source domain resources or models to serve the sentiment analysis tasks in the target domain, and it can effectively alleviate the problem of insufficient tagged data in the target domain. In this paper, we investigate deep transfer learning for text sentiment analysis, and focus on customer reviews for different domains. Given the shortcomings of the conventional deep neural network, such as convolutional neural network (CNN) cannot obtain the long-term dependence between features, recurrent neural network (RNN) cannot achieve parallel computing, we propose a novel neural network which joint dilated convolution and self-attention (ADC). The ADC takes residual learning as its basic framework, which can capture more distant information by introducing a dilation rate in the convolution. Self-attention is used to make the ADC pay more attention to important sentiment information. First, the ADC is pre-trained with the data of the source domain, and the weight of the pre-training is frozen. Then, the model is fine-tuned by the method of gradual unfreezing, and the weight is updated through the data of the target domain. Finally, the updated model can be utilized for the target domain sentiment analysis. By using customer review datasets, we carry out extensive experiments to demonstrate that this method has superior performance and can better achieve cross-domain sentiment analysis.

Multi Layered Rule-Based Technique for Explicit Aspect Extraction from Online Reviews

In the field of sentiment analysis, extracting aspects or opinion targets from user reviews about a product is a key task. Extracting the polarity of an opinion is much more useful if we also know the targeted Aspect or Feature. Rule based approaches, like dependency-based rules, are quite popular and effective for this purpose. However, they are heavily dependent on the authenticity of the employed parts-of-speech (POS) tagger and dependency parser. Another popular rule based approach is to use sequential rules, wherein the rules formulated by learning from the user’s behavior. However, in general, the sequential rule-based approaches have poor generalization capability. Moreover, existing approaches mostly consider an aspect as a noun or noun phrase, so these approaches are unable to extract verb aspects. In this article, we have proposed a multi-layered rule-based (ML-RB) technique using the syntactic dependency parser based rules along with some selective sequential rules in separate layers to extract noun aspects. Additionally, after rigorous analysis, we have also constructed rules for the extraction of verb aspects. These verb rules primarily based on the association between verb and opinion words. The proposed multi-layer technique compensates for the weaknesses of individual layers and yields improved results on two publicly available customer review datasets. The F1 score for both the datasets are 0.90 and 0.88, respectively, which are better than existing approaches. These improved results can be attributed to the application of sequential/syntactic rules in a layered manner as well as the capability to extract both noun and verb aspects.

Machine learning for assessing quality of service in the hospitality sector based on customer reviews

The increasing use of online hospitality platforms provides firsthand information about clients preferences, which are essential to improve hotel services and increase the quality of service perception. Customer reviews can be used to automatically extract the most relevant aspects of the quality of service for hospitality clientele. This paper proposes a framework for the assessment of the quality of service in the hospitality sector based on the exploitation of customer reviews through natural language processing and machine learning methods. The proposed framework automatically discovers the quality of service aspects relevant to hotel customers. Hotel reviews from Bogotá and Madrid are automatically scrapped from Booking.com. Semantic information is inferred through Latent Dirichlet Allocation and FastText, which allow representing text reviews as vectors. A dimensionality reduction technique is applied to visualise and interpret large amounts of customer reviews. Visualisations of the most important quality of service aspects are generated, allowing to qualitatively and quantitatively assess the quality of service. Results show that it is possible to automatically extract the main quality of service aspects perceived by customers from large customer review datasets. These findings could be used by hospitality managers to understand clients better and to improve the quality of service.

An Enhanced Hybrid Feature Selection Technique Using Term Frequency-Inverse Document Frequency and Support Vector Machine-Recursive Feature Elimination for Sentiment Classification

Sentiment classification is increasingly used to automatically identify a positive or negative sentiment in a text review. In classification, feature selection had always been a critical and challenging problem. Most of the related feature selection for sentiment classification techniques unable to overcome problems of evaluating the significant features that will reduce the classification performance. This paper proposes an enhanced hybrid feature selection technique to improve the sentiment classification based on machine learning approaches. First, two customer review datasets namely Sentiment Labelled and large IMDB are retrieved and pre-processed. Next, the proposed feature selection technique which is the hybridization of Term Frequency-Inverse Document Frequency (TF-IDF) and Supports Vector Machine (SVM-RFE) is developed and tested on these two datasets. TF-IDF aims to measure features importance. The SVM-RFE iteratively evaluates and ranks the features. For sentiment classification, only the ktop features from the ranked features will be used. Finally, the Support Vector Machine (SVM) classifier is deployed to observe the performance of the proposed technique. The performance is measured using accuracy, precision, recall, and F-measure. The experimental results show promising performances with 84.54% to 89.56% in the measurements especially from the large IMDB dataset. The results also outperformed other related techniques in certain datasets. Consequently, the proposed technique able to reduce from 19.25% to 70.5% of the features to be classified. This reduction rate is significant in optimally utilizing the computational resources while maintaining the efficiency of the classification performance.

Support Vector Machine (SVM) Optimization Using Grid Search and Unigram to Improve E-Commerce Review Accuracy

Electronic Commerce (E-Commerce) is distributing, buying, selling, and marketing goods and services over electronic systems such as the Internet, television, websites, and other computer networks. E-commerce platforms such as amazon.com and Lazada.co.id offer products with various price and quality. Sentiment analysis used to understand the product’s popularity based on customers’ reviews. There are some approaches in sentiment analysis including machine learning. The part of machine learning that focuses on text processing called text mining. One of the techniques in text mining is classification and Support Vector Machine (SVM) is one of the frequently used algorithms to perform classification. Feature and parameter selection in SVM significantly affecting the classification accuracy. In this study, we chose unigram as the feature extraction and grid search as parameter optimization to improve SVM classification accuracy. Two customer review datasets with different language are used which is Amazon reviews that written in English and Lazada reviews in the Indonesian language. 10-folds cross validation and confusion matrix are used to evaluating the experiment results. The experiment results show that applying unigram and grid search on SVM algorithm can improve Amazon review accuracy by 26,4% and Lazada reviews by 4,26%.

Robust product recommendation system using modified grey wolf optimizer and quantum inspired possibilistic fuzzy C-means

In recent years, several researchers have developed web-based product recommendation systems to assist customers in product search and selection during online shopping. In addition, the product recommendation systems deliver true personalization by recommending the products based on the other customer’s preferences. This study has investigated how the product recommendation system influences the customer’s decision effort and quality. In this study, the proposed system comprises of five major phases: data collection, pre-processing, key word extraction, keyword optimization and similar data clustering. The input data were collected from amazon customer review dataset. After the data collection, pre-processing was carried-out to enhance the quality of collected amazon data. The pre-processing phase comprises of two systems lemmatization and removal of stop-words & uniform resource locators (URLs). Then, a superior topic modelling method Latent Dirichlet allocation (LDA) along with modified grey wolf optimizer (MGWO) was applied in order to identify the optimal keywords. The extracted key-words were clustered into two forms (positive and negative) by applying a clustering algorithm named as quantum inspired possibilistic fuzzy C-means (QIPFCM). Experimental results showed that the proposed system achieved better performance in the product recommendation system compared to the existing systems in terms of accuracy, precision, recall and f-measure.

Graph based sentiment analysis using keyword rank based polarity assignment

Decision making by analyzing the underlying sentiment has become one of the challenging task with the explosion of rich source of user generated, diverse contents on the web. Sentiment analysis can ease the process of obtaining an overall sentiment by processing millions of reviews or documents altogether. A review or any textual data consists of numerous keywords, which hold different weightage based on various factors. An efficient ranking technique for those keywords is proposed in this paper in order to aid in the sentiment analysis process. A co-occurrence graph based statistical approach is adopted in this paper to find the global rank of the keywords. A novel node weighting technique is proposed, which will be used for the improvisation of the state of art method: Node and Edge Rank (NE-Rank) along with degree, to rank the keywords. The algorithm considers five different influential parameters to propose the node weighting technique. Also, a keyword may hold bi-polarity and depict different polarity according to the domain of application. Therefore, this paper proposes a novel, well organized and efficient sentiment analysis model using a graph based keyword ranking and domain specific rank based polarity assignment algorithm. The role or impact of an important keyword will be always more in comparison to a weaker one for determining the polarity of the review. Thus, the rank based polarity assignment technique is proposed with the use of the global ranks of keywords, to solve the domain dependency problem of the keywords. The proposed model is evaluated and validated using four different existing models for four different customer review datasets.

Aspect-based sentiment analysis using adaptive aspect-based lexicons

Reviews expressed in e-commerce websites have formed an important source of information for both consumers and enterprises. Text sentiment analysis approaches aim to detect the sentiments of written reviews in order to achieve a better understanding of public opinion towards entities. Aspect-based sentiment analysis deals with capturing sentiments expressed towards each aspect of entities. A common approach in sentiment analysis problems is to take advantage of lexicons to generate features for classification of reviews. Existing aspect-based approaches fail to properly adapt general lexicons to the context of aspect-based datasets which results in reduced performance. To address this problem, this paper proposes extensions of two lexicon generation methods for aspect-based problems; one using statistical methods, and another using a genetic algorithm presented in our previous works. The aforementioned lexicons are then fused with prominent static lexicons to classify the aspects in reviews; this outperforms our previous works according to the t-test results (with a p-value of less than 0.001). Experimental results indicate that the proposed approach outperforms baseline methods in aspect-based polarity classification on Bing Liu's customer review datasets and improves precision, recall and F-measure by 6.0, 1.0, and 7.4 percentage points respectively.

Exploring eWOM in online customer reviews: Sentiment analysis at a fine-grained level

Customer reviews in social media and electronic commerce Web sites contain valuable electronic word-of-mouth (eWOM) information of products, which facilitates firms’ business strategy and individual consumers’ comparison shopping. Exploring eWOM of products embedded in customer reviews has attracted interest from researchers in various fields. Coarse-grained and context-free sentiment analysis approaches have been used in existing researches, which however often fail to satisfy the firms’ demands of fine-grained extraction of market intelligence from social media. In this study, we propose an original method to explore eWOM of products based on sentiment analysis at fine-grained level from a large volume of online customer reviews. We illustrate a feature-based and context-sensitive sentiment analysis mechanism that can leverage the sheer volume of customer reviews in social media sites. A novel semi-supervised fuzzy product ontology mining algorithm is proposed to extract semantic knowledge from online customer reviews with positive or negative labels. Based on real-world online customer review data set, the proposed method shows remarkable performance improvement over baseline methods at exploring eWOM of product a fine-grained level. With the novel eWOM exploring method, firms can improve their product design and marketing strategies, and potential consumers can make better online purchase decisions.

Learning to recognize opinion targets using recurrent neural networks

Opinion target recognition can be deemed a problem of token-level sequence labelling, and each word in the sentence is assigned to a label with the standard BIO tagging scheme. Among a variety of methods, recently emerged recurrent neural network (RNN) is considered more effective to deal with such kinds of sequence annotation problem. Whereas existing RNN models mainly focus on learning the dependency relationships in the input sequence while ignoring the ones in the output sequence. To this end, we proposed an augmented RNN model called OLSRNN. The OLSRNN model adds self-connections to the output layer on the basics of conventional RNN models to further capture output temporal dependencies. Over the benchmark customer review datasets, experiment results demonstrate the effectiveness of the proposed approach in opinion target recognition in comparison with other baseline methods.

Product Feature Ranking and Popularity Model based on Sentiment Comments

This paper proposes the development of a model to determine feature popularity ranking for products in the market. Each feature that is reviewed by a customer has a relation to sentiment words present in the sentences within a customer review. Feature quantity of a product, derived from customer review dataset, cannot be used as a benchmark to determine customers’ preferences since each feature is influenced by sentiment words that give it either a positive or negative meaning. A positive meaning shows that the feature is liked by user; and a negative meaning shows that it is disliked by user. This study finds that sentiment assessments by users play an important role in determining feature popularity ranking; and they affect the feature of a product. Thus, this study proposes the development of a model that takes into account the importance of sentiment assessments present in each sentence within a customer review of a product feature. A case study has been conducted in proving that the developed model is able to produce a list of product feature popularity ranking. Results of this experimental model is also put into simple comparative analysis with a few models from previous studies.

Effective Sentiment Analysis for Opinion Mining Using Artificial Bee Colony Optimization

Opinions play important role in the process of knowledge discovery or information retrieval and can be considered as a sub discipline of Data Mining. The huge quantity of information on web platforms put together feasible for exercise as data sources, in applications based on opinion mining and classification. An effective sentiment analysis process proposes in this research for mining and classifying the opinions. The phases of the proposed research are: (1) Data Pre-processing Phase (2) Potential Feature Extraction Phase (3) Opinion Extraction and Mining Phase and (4) Opinion Classification Phase. Initially, the datasets from various web documents get preprocessed and gives as part-of-speech tagged text. An Improved High Adjective Count (IHAC) Algorithm employs on the Part-Of-Speech tagged text to extract the potential features. Improved High Adjective Count Algorithm effectively optimizes the scores of the nouns to extract the potential features. An Artificial Bee Colony (ABC) Algorithm works under the IHAC algorithm for providing opinion scores and also for giving ranks for every noun. Max Opinion Score Algorithm can be then helpful to extract the opinion words followed by the classification phase, in which, ID3 algorithm utilizes to classify the review into three kinds positive, negative and neutral based on the opinions. The implementation is carried out on Customer Review Datasets and Additional Review Datasets with the aid of JAVA platform and also the experimentation results are analyzed.

Automatic multi-way domain concept hierarchy construction from customer reviews

Abstract A concept hierarchy is important for many applications to manage and analyze text corpora. In the literature, most previous hierarchy construction works are under the assumption that the semantic relations in the concept hierarchy can be extracted from a text corpus, which is not fully satisfied for short and informal texts, e.g. tweets and customer reviews. And many works utilize hierarchical clustering methods to get the final concept hierarchy, in which the resulting binary-tree form concept hierarchy cannot fit the demand in many applications. In this paper, we propose a general process for building a concept hierarchy from customer reviews with an appropriate depth. The process can be divided into three steps. First, all highly ranked topic words are extracted as concept words using a topic model. And a word sense disambiguation task is performed to derive the possible semantics of the words. Then, the distances between these words are computed by combining their contexts and relations in the WordNet. Finally, all words are organized using a modified multi-way hierarchical clustering method. In addition, a new concept hierarchy evaluation model is presented. Our approach is compared to approaches using hierarchical clustering methods on the Amazon Customer Review data set, and the results show that our approach can get higher similarity scores with the reference concept hierarchy.

WordNet Based Multi-Way Concept Hierarchy Construction from Text Corpus

In this paper, we propose an approach to build a multi-way concept hierarchy from a text corpus, which is based on WordNet and multi-way hierarchical clustering. In addition, a new evaluation metric is presented, and our approach is compared with 4 kinds of existing methods on the Amazon Customer Review data set.