Product Feature Extraction Research Articles

Ranking products through online opinions: A text analysis and regret theory-based approach

With the development of e-shopping, a list of similar products can be found with a large volume of valuable customer reviews online. However, it is generally difficult to compare various aspects of similar products effectively by understanding all relevant online opinions. To help consumers, in this study, how products are ranked according to online reviews is investigated. Firstly, an SC-LDA (Seed Constraint-Latent Dirichlet Allocation) model, which is an extension of the classical topic model LDA (Latent Dirichlet Allocation), is proposed to extract product features. The must-link and cannot-link seed constraints are invited to estimate the probability expansion/reduction value. They help to affect the topic allocation by additional constraints in Gibbs sampling for a higher accuracy on feature extraction. Secondly, an improved convolutional memory neural network model is devised to analyze the sentiment polarity. It takes the advantages of CNN (convolutional neural network) and Bi-LSTM (bidirectional Long Short-Term Memory) and performs dynamic pooling in CNN to prevent the loss of important features. Besides, the concept of group satisfaction degree is introduced, which makes products be compared according to the Regret Theory. It ranks products without a commonly applied reference point and take consumer psychology into considerations. Finally, in the case study, an illustrative example is presented to evaluate the proposed framework. Categories of experiments show that the proposed framework provides consumers with effective purchase suggestions. Code metadataPermanent link to reproducible Capsule: https://doi.org/10.24433/CO.6445683.v1 and https://doi.org/10.24433/CO.2658577.v1.

A Deep Learning-Based Sentiment Analysis Approach for Online Product Ranking With Probabilistic Linguistic Term Sets

The probabilities linguistic term set (PLTS) is an efficient tool to represent sentimental intensities hidden in unstructured text reviews that are useful for multicriteria online product ranking. Traditional machine learning-based sentiment analysis methods adopted in existing studies to obtain PLTSs often result in unsatisfying prediction accuracy and, thus, inevitably affect product ranking results. To overcome this limitation, in this study, we propose a deep learning-based sentiment analysis approach to produce PLTSs from online product reviews to rank online products. A natural language processing-based method is first applied to extract product features and corresponding feature texts from online reviews. Then, state-of-the-art deep learning-based models are implemented to conduct the sentiment classification for online product/feature review texts. To ensure classification accuracy, we propose an experimental matching mechanism to identify the level of sentiment tendency for all rating labels of a review dataset and then match each label with the most appropriate linguistic term. The experimental results reveal that our matching mechanism can benefit the training of a text classification model to identify sentiment tendencies from review texts with high prediction accuracy and with the help of the trained classification model, our approach can predict sentimental intensities of the extracted features' texts in the form of PLTSs with competitive accuracy. A case study of applying PLTSs output from our approach to an online product decision-making problem is also provided to validate the applicability of our approach.

Product feature sentiment analysis based on GRU-CAP considering Chinese sarcasm recognition

Sentiment analysis based on product online reviews is an important measure in grasping customer’s satisfaction with products, which has the problem of inadequate feature extraction and difficulty in identifying the sentiment polarity of texts with irony. To solve the problem, this paper proposes a novel method of product feature sentiment analysis based on gate recurrent units and capsule neural network (GRU-CAP). Firstly, product online reviews are preprocessed to build a review data set. The feature extraction model of GRU-CAP is then used to extract the product features contained in the data set and generate the corresponding product feature index. After combining the feature index with the data set to form new input data, the recognition of positive/negative emotions, sarcastic/non-sarcastic semantics is carried out through the sentiment analysis model of GRU-CAP. Finally, the recognition results are integrated to achieve fine-grained product feature sentiment analysis. The effectiveness and practicability of the method is validated by an example product-mobile phone on an e-commerce platform. The results show that the method can extract product features and identify sentiment polarity from product online review texts with irony more accurately, so as to provide better product design support for enhancing customer satisfaction with products more effectively.

Multivariate SVR Demand Forecasting for Beauty Products Based on Online Reviews

Owing to changes in consumer attitudes, the beauty consumer population is growing rapidly and the demands of beauty consumers are variable. With a wide range of beauty products and exaggerated product promotions, consumers rely more on online reviews to perceive product information. In this paper, we propose a demand forecasting model that takes into account both product features and product emotional needs based on online reviews to help companies better develop production and sales plans. Firstly, a Word2vec model and sentiment analysis method based on a sentiment dictionary are used to extract product features and factors influencing product sentiment; secondly, a multivariate Support Vector Regression (SVR) demand prediction model is constructed and the model parameters are optimized using particle swarm optimization; and finally, an example analysis is conducted with beauty product Z. The results show that compared with the univariate SVR model and the multivariate SVR model with only product feature demand as the influencing factor, the multivariate SVR model with both product feature and product sentiment demand as influencing factors has a smaller prediction error, which can enable beauty retail enterprises to better grasp consumer demand dynamics, make flexible production and sales plans, and effectively reduce production costs.

Product feature extraction from Chinese online reviews: application to product improvement

Online product reviews are valuable resources to collect customer preferences for product improvement. To retrieve consumer preferences, it is important to automatically extract product features from online reviews. However, product feature extraction from Chinese online reviews is challenging due to the particularity of the Chinese language. This research focuses on how to accurately extract and prioritize product features and how to establish product improvement strategies based on the extracted product features. First, an ensemble deep learning based model (EDLM) is proposed to extract and classify product features from Chinese online reviews. Second, conjoint analysis is conducted to calculate the corresponding weight of each product feature and a weight-based Kano model (WKM) is proposed to classify and prioritize product features. Various comparative experiments show that the EDLM model achieves impressive results in product feature extraction and outperforms existing state-of-the-art models used for Chinese online reviews. Moreover, this study can help product managers select the product features that have significant impact on enhancing customer satisfaction and improve products accordingly.

Sentiment Analysis using Latent Dirichlet Allocation for Aspect Term Extraction

This work proposes a sentiment analysis approach for decision-making in product design, analysis, and market share. The approach incorporates user-generated text data in the form of consumer reviews to extract product features using topic-based modeling methods. Latent Dirichlet Allocation (LDA) is employed to extract aspect categories from the data and identify the sentiment of each review using the VADER sentiment analyzer. The performance of the proposed method is evaluated in terms of accuracy, with an achieved result of 80%. The extracted topics are also summarized to provide leads for product design and quality assurance. The approach can be used by manufacturers, retailers, and suppliers to understand customers' opinions about their products better and make better decisions. LDA is a powerful unsupervised method that can extract latent topics from a collection of documents; this method has been widely used in text mining, information retrieval, and natural language processing. The accuracy can be improved by using more sophisticated models or more data.

Fake Review Identification Methods Based on Multidimensional Feature Engineering

Product reviews in electronic platforms are very valuable to potential customers, product manufacturers, and product sellers. Their data contain huge business opportunities. Therefore, this paper analyzes the views, attitudes, and emotions expressed in these reviews. It presents three fake review identification methods based on multidimensional feature engineering. Under the premise of adding product feature extraction and opinion sentence judgment, six feature parameters are defined to identify fake reviews, and a fake review identification model based on multidimensional feature engineering is constructed. Then, the effectiveness of the selected feature engineering is verified. Based on the multidimensional feature engineering model, a fake review identification algorithm based on multidimensional feature engineering of union relationship, an identification algorithm based on weighted multidimensional feature engineering scoring, and an identification algorithm based on weighted multidimensional feature engineering classification are proposed. The execution effects of the three methods are compared. Fake review identification models based on multidimensional feature engineering can effectively filter fake reviews.

Feature extraction of search product based on multi-feature fusion-oriented to Chinese online reviews

The increasing Chinese online reviews contain rich product demand information, especially for search products. This study suggests a product feature extraction model from online reviews based on multi-feature fusion named PFEMF (products features extraction based on multi-feature fusion) model. Combining sentence and word characteristics of Chinese online reviews, the model explores the lexical features, frequency features, span features, and semantic similarity features of words. And then, they are fused to identify the features that customers are concerned about most by sequential relationship analysis. The identified product feature provides direction for product innovation and facilitates the product selection for customers. Finally, the study takes iPad Air as an example to prove this model. The results show that the extraction performance of the PFEMF model is superior to the traditional term frequency-inverse document frequency (tf-idf) algorithm, word span algorithm, and semantic similarity algorithm.

Review Mining for Experiential Products Incorporating Word2vec and Review Sentiment Tendencies

In view of the difficulty in mining the characteristics of experience-based products, this paper proposes a method combining LDA and Word2vec to extract product features. This method uses the emotion dictionary to carry out fine-grained analysis on these features, which improves the defect that the effect of topic clustering is not obvious. The result shows that it can effectively mine the feature words with high relevance, and put forward more targeted suggestions for producers.

Algorithm for extracting product feature from e-commerce comment

<span>Reviews of e-commerce play an important role in online purchasing decisions. Consumers are likely to read reviews and comments on products from other consumers. In addition to those opinions that reflect consumers' trust in products, it also provides each product's distinctive properties. Today, there are many online reviews, resulting in enormous comments and suggestions. However, as fully reading reviews is quite difficult, this article presents 3 algorithms for automatic extraction of product features hidden in e-commerce reviews: a traditional frequency-based product feature extraction (F-PFE), syntax analyzer system (SAS), and the hybrid approach called the frequency and syntax-based product feature extraction (FaS-PFE). The proposed algorithms were tested against 4 different types of products: shampoo, skincare, mobile phone, and tablet, using reviews from amazon.com. Based on the product review used in this study, it was found that the SAS can help improve the performance in terms of precision by 15% when compared with the traditional F-PEE approach. When considering both the word frequency and syntax, FaS-PFE clearly outperforms the other two approaches with 94.00% precision and 95.13% recall.</span>

Product and design feature-based similar process retrieval and modeling for mold manufacturing

The design, machining, and assembly processes used to manufacture molds are knowledge-intensive, costly, and time-consuming operations requiring considerable human resources. Mold manufacturing is a typical make-to-order system. The characteristics of new products (molds or products produced using molds) determine the process and resource plan for mold manufacturing. Therefore, it is essential to store and inherit knowledge regarding products or mold design and process planning. In this study, we extract product features from images of products. We measure the similarity between features of new products and those of products previously manufactured. The clustering method is used to search for similar processes and resources among previously manufactured products. Moreover, design characteristics such as mold type, structure, and size, obtained from design data, are used to support this search for similar processes and resources. Herein, we propose a framework for a similarity retrieval model based on results of the product and design feature extraction. Convolutional neural networks and an agglomerative hierarchical clustering algorithm are used to derive similar processes. Furthermore, we propose a semiautomatic procedure to construct a collaborative process model that includes product, process, and resource information. This collaborative process modeling supports the visualization and analyses of complex processes for new products. Through this study, we endeavor to construct a collaborative process model by formulating it as an optimization problem.

Building of Visual Analysis System for Design of Youth Sports Shoe Products Based on Comment Mining

This paper proposes a visual analysis system for data collection, pre-preprocessing and sentiment analysis to support whole-process innovative design of youth sports shoes. This system aims to analyze the sales data of sports shoe e-commerce sellers before and after design, provide reference for key design elements in innovative design of sports shoes based on 3D printing and realize dynamic closed design of pre-design data investigation and analysis and post-design feedback data analysis. With dynamic data collection technology based on a crawler, product feature extraction model, sentiment analysis model and big data visualization technology, this system can better satisfy data analysis requirement for innovative design of sports shoes.

A Deep Learning Method for Yogurt Preferences Prediction Using Sensory Attributes

During the development of innovative products, consumer preferences are the essential factors for yogurt producers to improve their market share. A high-performance prediction method will be beneficial to understand the intrinsic relevance between preferences and sensory attributes. In this study, a novel deep learning method is proposed that uses an autoencoder to extract product features from the sensory attributes scored by experts, and the sensory features acquired are regressed on consumer preferences with support vector machine analysis. Model performance analysis, hedonic contour mapping, and feature clustering were implemented to validate the overall learning process. The results showed that the deep learning model can vouch an acceptable level of accuracy, and the hedonic mapping reflected could supply a great help for producers’ product design or modification. Finally, hierarchical clustering analysis revealed that for all three brands of yogurts, low temperature (4 °C) storage for no more than 4 weeks can promise the highest consumer preferences.

Processes and methods of information fusion for ranking products based on online reviews: An overview

Over the past few years, more and more consumers have come to read online reviews when they shop online. To support consumers' purchase decisions, many scholars focus on ranking products based on online reviews and propose various methods and techniques. Generally, the process of information fusion for ranking products based on online reviews consists of three stages: product feature extraction, sentiment analysis, and ranking products. In this paper, we review the existing studies on processes and methods of information fusion for each stage. Furthermore, we briefly review the existing research on information fusion based on online reviews in other fields. Finally, we summarize the main conclusions of this paper and point out the future research direction.

Supporting the construction of affective product taxonomies from online customer reviews: an affective-semantic approach

ABSTRACTConsumers today not only consider the functionality and reliability of products, but also concern with affective aspects of products to meet their emotional needs. Products with good affective design can excite consumers’ psychological feelings and enhance consumer satisfaction. Affective engineering aims to discover relationships between product features and affective preferences for affective design. Traditional methods rely heavily on manual surveys, which are costly, and the affective design knowledge is difficult to share and update. There is a need to develop an efficient way to build a common knowledge representation. In this paper, we propose an affective-semantic approach to automatically construct affective product taxonomy based on online consumer reviews. We incorporate affective engineering and semantic analysis to extract product features and affective attributes from online product information. We construct taxonomy by relating the extracted product features and affective attributes based on their meaning. To evaluate the effectiveness of the approach, experiments have been conducted using public available data. The results showed that the approach can effectively identify and measure affective information. It could help develop a common understanding of the design domain for reuse and expansion. It could also assist product designers review existing products based on affective aspects and consumer perspectives.

Product Quality Assessment Based on Online Reviews

Reviews can reflect the degree of consumers' satisfaction and views on product quality, and consumers tend to read product reviews and then get helpful information about product quality before placing an order in e-commerce platforms. However, the existing research mainly focus on the assessment of review quality, fake review detection, opinion mining, and there is little research to assess product quality from the perspectives of product features based on reviews objectively and quantifialy. Therefore, the authors propose a method to assess product quality based on reviews in a granularity of product feature. The authors define the related quality dimensions and develop the corresponding assessment models, assess the review quality crawled from an e-commerce platform, then extract product features and opinion words from the quality reviews, and finally assess product quality on the extracted and consumer-concerned features. Experiment results demonstrate the methodology can achieve the assessment of product quality on any feature objectively and quantificationally.

Multiple affective attribute classification of online customer product reviews: A heuristic deep learning method for supporting Kansei engineering

Kansei engineering is a product design methodology that takes into account the affective needs of customers and translates them into product parameters. Traditionally, it uses questionnaire survey to collect and understand affective responses of customers. However, this process requires active involvement of participants, which is expensive, time-consuming, knowledge intensive and labour intensive. Therefore, it is only suitable for relatively small-scale operations. Today, due to the rapid development of e-commerce and social media, customers often write reviews to describe their feelings and comments after purchasing a product. A larger number of customer product reviews are available on the Internet, providing free, trustworthy, and growing data. Recently, many studies have used opinion mining and sentiment analysis to analyse customer product reviews. However, they focus primarily on product feature extraction or classification from a single perspective (i.e. good or bad). Research on multiple affective attributes has received less attention. In this paper, we propose a heuristic deep learning method that extracts affective opinions from customer product reviews, and classify them into seven pairs of affective attributes (i.e. like–dislike, aesthetic–inaesthetic, soft–hard, small–big, useful–useless, reliable–unreliable, and recommended–not recommended). An experiment has been conducted by using the product reviews available in Amazon.com. The results show that the accuracy of the proposed method is above 86%, which outperforms the baseline methods. The extracted information has the potential to help both customers and product designers understand the customer comments from an affective perspective and assist them in making decisions.

Application of Sentiment Analysis on Product Review E-Commerce

The lack of buying product via online are the consumer cannot touch, try, or even see it directly. Then how does the consumer believe the product they like, is the correct product to be bought. The main key is product review from the consumer who have bought and try the product. The more number of product review on certain product or the popular product caused difficulties for the consumer to decide which product they should choose. For that we require a solution, which is the writer will apply sentiment analysis to classify every product review into positive orientation, negative or neutral and also produce summary of product review based on product feature to help reading process, product review and decision making. The writer step is (1) data collecting and preprocessing, (2) product feature extraction using Double Propagation, (3) deciding sentiment orientation, (4) classifying by using Naïve Bayes Classifier and Support Vector Machine, and (5) summary generation based on product feature. These stages run in a simulator and the result of the classification from both methods compared.

A Decision Method for Online Purchases Considering Dynamic Information Preference Based on Sentiment Orientation Classification and Discrete DIFWA Operators

Online reviews are crucial for evaluating product features and supporting consumers' purchase decisions. However, as a result of online buying behaviors, consumer habits, and discrete dynamic distribution characteristics of online reviews, and consumers typically randomly choose a limited number of reviews from discrete time frames among all reviews and give more weight to recent review information and less weight to earlier information to support their online purchase decisions; moreover, existing studies have ignored the discrete random dynamic characteristics and dynamic information preferences of consumers. To address this issue, this paper proposes a method based on sentiment orientation classification and discrete DIFWA (DDIFWA) operators for online purchase decisions considering dynamic information preferences. In this method, we transformed review texts from original discrete time slices to discrete random features, extracted product features based on the constructed feature and sentiment dictionaries, and matched pairs of features and sentiment phrases in the dictionaries. We subsequently employed three types of semantic orientation by defining semantic rules to extract the product features of each review. Of note, the semantic orientations were transformed from discrete time to semantic intuitionistic fuzzy numbers and semantic intuitionistic fuzzy information matrixes. Furthermore, we proposed two DDIFWA operators to aggregate the dynamic semantic intuitionistic fuzzy information. Specifically, we obtained the rankings of alternative products and their features to support consumers' purchase decisions using the intuitionistic fuzzy scoring function and the “vertical projection distance” method. Finally, comparisons and experiments are provided to demonstrate the plausibility of our methods.

Analysis on Opinion Words Extraction in Electronic Product Reviews

Online user reviews are increasingly becoming important for measuring the quality of different products and services. Sentiment classification or opinion mining involves studying and building a system that collects data from online and examines the opinions. Sentiment classification is also defined as opinion extraction as the computational research area of subjective information towards different products. Opinion mining or sentiment classification has attracted in many research areas because of its usefulness in natural language processing and other area of applications. Extracting opinion words and product features are also important tasks in opinion mining. In this work an unsupervised approach was proposed to extract opinions and product features without training examples. To obtain the dependency relation between the product aspects and opinions, this work used StanfordCoreNLP dependency parser. From these relations, rules are predified to extract product and opinions. The main advantage of this approach is that there is no need for training data and it has domain independence. Acoording to the experimental results, the modified algorithm gets better results than the double propagation algorithm.