• Journal of KIISE:Databases
• /
• v.36 no.4
• /
• pp.254-262
• /
• 2009

As the trend of e-business activities develop, customers come into contact with products through on-line shopping sites and lots of customers refer product reviews before the purchasing on-line. However, as the volume of product reviews grow, it takes a great deal of time and effort for customers to read and evaluate voluminous product reviews. Lately, attention is being paid to Opinion Mining(OM) as one of the effective solutions to this problem. In this paper, we propose an efficient method for opinion sentiment classification of product reviews using product specific context information of words occurred in the reviews. We define the context information of words and propose the application of context for sentiment classification and we show the performance of our method through the experiments. Additionally, in case of word corpus construction, we propose the method to construct word corpus automatically using the review texts and review scores in order to prevent traditional manual process. In consequence, we can easily get exact sentiment polarities of opinion words in product reviews.

• ETRI Journal
• /
• v.46 no.4
• /
• pp.633-647
• /
• 2024

In e-commerce platforms, sentiment analysis on an enormous number of user reviews efficiently enhances user satisfaction. In this article, an automated product recommendation system is developed based on machine and deep-learning models. In the initial step, the text data are acquired from the Amazon Product Reviews dataset, which includes 60 000 customer reviews with 14 806 neutral reviews, 19 567 negative reviews, and 25 627 positive reviews. Further, the text data denoising is carried out using techniques such as stop word removal, stemming, segregation, lemmatization, and tokenization. Removing stop-words (duplicate and inconsistent text) and other denoising techniques improves the classification performance and decreases the training time of the model. Next, vectorization is accomplished utilizing the term frequency-inverse document frequency technique, which converts denoised text to numerical vectors for faster code execution. The obtained feature vectors are given to the modified convolutional neural network model for sentiment analysis on e-commerce platforms. The empirical result shows that the proposed model obtained a mean accuracy of 97.40% on the APR dataset.

• Culinary science and hospitality research
• /
• v.22 no.8
• /
• pp.109-121
• /
• 2016

This study analyzed the effect of five sub-variables' attribute of HMR: features of information, diversity, promptness, price and convenience, on the consumer purchasing intention. In addition, the regulation effect of positive reviews and negative reviews of consumers' online reviews between HMR selection attribute and purchasing intention was also tested. Results are following. First, convenience feature (B=.577, p<.001) and diversity feature (B=.093, p<.01) among the effect of HMR selection attribute had a positive (+) effect on purchasing intention. On the other hand, promptness feature (B=.235, p<.001) and price feature (B=.161, p<.001), and information feature (B=.288, p<.001) were not significant effect on purchasing intention. Second, result of regulation effect of the positive reviews of consumer's online review between the selection attribute of the HMR product and consumers' purchasing intention, in the first-stage model in which the selection attribute of the HMR product is input as an independent variable, there was a significant positive (+) effect on all the features of convenience, diversity, promptness, price, and information. In addition, there was significant positive (+) main effect (B=.472, p<.001) in the second step model in which the consumers' positive reviews, that is a regulation variable. Furthermore, the feature of price (B=.068, p<.05) had a significant positive (+) effect in the third stage in which the selection attribute of the HMR product that is an independent variable and the interaction of the positive review. However, the feature of information (B=-.063, p<.05) showed negative (-) effect, and there was no effect on the features of convenience, diversity, and promptness. Third, as a result of testing the regulation effect of the negative reviews of consumers' online reviews between HMR product selection attribute and consumers' purchasing intention, in the first-stage model in which the selection attribute of the HMR product was a positive (+) effect on all the features of convenience, diversity, promptness, price, and information. In the second-stage model in which consumers' negative reviews (B=-.113, p<.001) had negative (-) effect. In the third-stage in which the selection attribute of the HMR product and the interactions of the negative reviews was a positive (+) effect with the feature of price (B=.113, p<.01). Last, there was no effect at all on the features of convenience, promptness, and information.

• Asia pacific journal of information systems
• /
• v.25 no.4
• /
• pp.784-804
• /
• 2015

This study investigates the effects of a two-sided message on product attitude and purchase intention by using a message structure variable, such as attribute importance in the context of online consumer reviews (OCRs). Study 1 explains the previous inconsistent results of a two-side message by comparing a one-side message and a two-side message by using the attribute importance in negative reviews. Study 2 determines the reasons for the inconsistent results of a refutational two-sided message research by using the attribute importance in negative reviews and website trust. Two experiments are designed to test our hypotheses. The first experiment is a $2{\times}2$ factorial design with 84 participants. The second experiment uses a $2{\times}2{\times}2$ factorial design with 196 participants. In study 1, two-sided OCRs are more credible than one-sided OCRs, and two-sided OCRs that use low important attributes are more effective in making favorable product attitude/purchase intention. In study 2, refutational two-sided OCRs that use high attribute importance render positive effects on product attitudes in trustworthy websites. However, the refutation could negatively affect product attitude/purchase intention in low trustworthy websites.

• The KIPS Transactions:PartB
• /
• v.16B no.6
• /
• pp.497-502
• /
• 2009

In this paper, we introduce an automatic product feature extracting system that improves the efficiency of product review analysis. Our system consists of 2 parts: a review collection and correction part and a product feature extraction part. The former part collects reviews from internet shopping malls and revises spoken style or ungrammatical sentences. In the latter part, product features that mean items that can be used as evaluation criteria like 'size' and 'style' for a skirt are automatically extracted by utilizing term statistics in reviews and web documents on the Internet. We choose nouns in reviews as candidates for product features, and calculate degree of association between candidate nouns and products by combining inner association degree and outer association degree. Inner association degree is calculated from noun frequency in reviews and outer association degree is calculated from co-occurrence frequency of a candidate noun and a product name in web documents. In evaluation results, our extraction method showed an average recall of 90%, which is better than the results of previous approaches.

• The Journal of the Korea Contents Association
• /
• v.20 no.7
• /
• pp.316-330
• /
• 2020

The study analyzes reviews of hardware products, customer service products, and products that take the form of a convergence of hardware and cloud services in ICT using text mining. We derive keywords of each review and find the differentiation of words that are used to derive topics. A cluster analysis is performed to categorize reviews into their respective clusters. Through this study, we observed which keywords are most often used for each product type and found topics that express the characteristics of products and services using topic modeling. We derived keywords such as "professional" and "technician" which are topics that suggest the excellence of the service provider in the review of service products. Further, we identified adjectives with positive connotations such as "favorite", "fine", "fun", "nice", "smart", "unlimited", and "useful" from Amazon Eco review, an integrated product and service. Using the cluster analysis, the entire review was clustered into three groups, and three product type reviews exclusively resulted in belonging to each different cluster. The study analyzed the differences whereby consumer needs are expressed differently in reviews depending on the type of product and suggested that it is necessary to differentiate product planning and marketing promotion according to the product type in practice.

• 한국HCI학회:학술대회논문집
• /
• 2009.02a
• /
• pp.660-663
• /
• 2009

Consumers express how they evaluate what they purchased by writing reviews especially when they purchased products online. By analyzing the reviews about a product, it will be possible to find out what the consumers liked and disliked about the product. It will be also possible to identify the general consensus on what matters in purchaing certain product type such as a laptop if many reviews about many instances of a particular product type is analyzed. However, it takes a lot of time to manually analyzing the reviews. Thus, we propose to use two natural language processing oriented computational techniques to analyze a large number of reviews. The techniques are text classification and information extraction. We developed an review analysis system and conducted experiments against the reviews about the laptop computers posted on the Naver information portal.

• Journal of Information Technology Services
• /
• v.21 no.4
• /
• pp.75-90
• /
• 2022

From the past to the present, reviews have had much influence on consumers' purchasing decisions. Companies are making various efforts, such as introducing a review incentive system to increase the number of reviews. Recently, as various types of reviews can be left, reviews have begun to be recognized as interesting new content. This way, reviews have become essential in creating loyal customers. Therefore, research and utilization of reviews are being actively conducted. Some studies analyze reviews to discover customers' needs, studies that upgrade recommendation systems using reviews, and studies that analyze consumers' emotions and attitudes through reviews. However, research that predicts the future using reviews is insufficient. This study used a dataset consisting of two reviews written in pairs with differences in usage periods. In this study, the direction of consumer product evaluation is predicted using KoBERT, which shows excellent performance in Text Deep Learning. We used 7,233 reviews collected to demonstrate the excellence of the proposed model. As a result, the proposed model using the review text and the star rating showed excellent performance compared to the baseline that follows the majority voting.

• The Journal of Society for e-Business Studies
• /
• v.16 no.4
• /
• pp.125-137
• /
• 2011

Product reviews in online shopping sites can serve as a useful guideline to buying decisions of customers. However, due to the massive amount of such reviews, it is almost impossible for users to read all the product reviews. For this reason, e-commerce sites provide users with useful reviews or statistics of ratings on products that are manually chosen or calculated. Opinion mining or sentiment analysis is a study on automating above process that involves firstly analyzing users' reviews on a product to tell if a review contains positive or negative feedback, and secondly, providing a summarized report of users' opinions. Previous researches focus on either providing polarity of a user's opinion or summarizing user's opinion on a feature of a product that result in relatively low usage of information that a user review contains. Actual user reviews contains not only mere assessment of a product, but also dissatisfaction and flaws of a product that a user experiences. There are increasing needs for effective analysis on such criteria to help users on their decision-making process. This paper proposes a model that stores various types of user reviews in a data warehouse, and analyzes integrated reviews dynamically. Also, we analyze reviews of an online application shopping site with the proposed model.

• Asia pacific journal of information systems
• /
• v.31 no.4
• /
• pp.472-490
• /
• 2021

The detection of products with manipulated reviews has received widespread research attention, given that a truthful, informative, and useful review helps to significantly lower the search effort and cost for potential customers. This study proposes a method to recognize products with manipulated online customer reviews by examining the sequence of each review's sentiment, readability, and rating scores by product on randomness, considering the example of a Russian online retail site. Additionally, this study aims to examine the association between brand awareness and existing manipulation with products' reviews. Therefore, we investigated the difference between well-known and unknown brands' products online reviews with and without manipulated reviews based on the average star rating and the extremely positive sentiment scores. Consequently, machine learning techniques for predicting products are tested with manipulated reviews to determine a more useful one. It was found that about 20% of all product reviews are manipulated. Among the products with manipulated reviews, 44% are products of well-known brands, and 56% from unknown brands, with the highest prediction performance on deep neural network.