• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.2
• /
• pp.263-283
• /
• 2024

Users' comments after online shopping are critical to product reputation and business improvement. These comments, sometimes known as e-commerce reviews, influence other customers' purchasing decisions. To confront large amounts of e-commerce reviews, automatic analysis based on machine learning and deep learning draws more and more attention. A core task therein is sentiment analysis. However, the e-commerce reviews exhibit the following characteristics: (1) inconsistency between comment content and the star rating; (2) a large number of unlabeled data, i.e., comments without a star rating, and (3) the data imbalance caused by the sparse negative comments. This paper employs Bidirectional Encoder Representation from Transformers (BERT), one of the best natural language processing models, as the base model. According to the above data characteristics, we propose the F_MixBERT framework, to more effectively use inconsistently low-quality and unlabeled data and resolve the problem of data imbalance. In the framework, the proposed MixBERT incorporates the MixMatch approach into BERT's high-dimensional vectors to train the unlabeled and low-quality data with generated pseudo labels. Meanwhile, data imbalance is resolved by Focal loss, which penalizes the contribution of large-scale data and easily-identifiable data to total loss. Comparative experiments demonstrate that the proposed framework outperforms BERT and MixBERT for sentiment analysis of e-commerce comments.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.20 no.2
• /
• pp.202-207
• /
• 2010

In this paper, we present an automatic sentiment classification method for on-line movie reviews that do not contain explicit sentiment rating scores. For the sentiment polarity classification, positive or negative, we use a Support Vector Machine classifier based on syllable kernel that is an extended model of string kernel. We give some experimental results which show that proposed syllable kernel model can be effectively used in sentiment classification tasks for on-line movie reviews that usually contain a lot of grammatical errors such as spacing or spelling errors.

• Journal of Information Science Theory and Practice
• /
• v.3 no.1
• /
• pp.6-23
• /
• 2015

This study develops an effective method for sentiment analysis of user-generated content on drug review websites, which has not been investigated extensively compared to other general domains, such as product reviews. A clause-level sentiment analysis algorithm is developed since each sentence can contain multiple clauses discussing multiple aspects of a drug. The method adopts a pure linguistic approach of computing the sentiment orientation (positive, negative, or neutral) of a clause from the prior sentiment scores assigned to words, taking into consideration the grammatical relations and semantic annotation (such as disorder terms) of words in the clause. Experiment results with 2,700 clauses show the effectiveness of the proposed approach, and it performed significantly better than the baseline approaches using a machine learning approach. Various challenging issues were identified and discussed through error analysis. The application of the proposed sentiment analysis approach will be useful not only for patients, but also for drug makers and clinicians to obtain valuable summaries of public opinion. Since sentiment analysis is domain specific, domain knowledge in drug reviews is incorporated into the sentiment analysis algorithm to provide more accurate analysis. In particular, MetaMap is used to map various health and medical terms (such as disease and drug names) to semantic types in the Unified Medical Language System (UMLS) Semantic Network.

• Journal of Korea Trade
• /
• v.25 no.2
• /
• pp.22-45
• /
• 2021

Purpose - This study aims to identify new attributes by analyzing reviews conducted by global automaker customers and to examine the influence of these attributes on satisfaction ratings in the U.S. automobile sales market. The present study used J.D. Power for customer responses, which is the largest online review site in the USA. Design/methodology - Automobile customer reviews are valid data available to analyze the brand personality of the automaker. This study collected 2,998 survey responses from automobile companies in the U.S. automobile sales market. Keyword analysis, topic modeling, and the multiple regression analysis were used to analyze the data. Findings - Using topic modeling, the author analyzed 2,998 responses of the U.S. automobile brands. As a result, Topic 1 (Competence), Topic 5 (Sincerity), and Topic 6 (Prestige) attributes had positive effects, and Topic 2 (Sophistication) had a negative effect on overall customer responses. Topic 4 (Conspicuousness) did not have any statistical effect on this research. Topic 1, Topic 5, and Topic 6 factors also show the importance of buying factors. This present study has contributed to identifying a new attribute, personality. These findings will help global automakers better understand the impacts of Topic 1, Topic 5, and Topic 6 on purchasing a car. Originality/value - Contrary to a traditional approach to brand analysis using questionnaire survey methods, this study analyzed customer reviews using text mining. This study is timely research since a big data analysis is employed in order to identify direct responses to customers in the future.

• Culinary science and hospitality research
• /
• v.22 no.6
• /
• pp.61-70
• /
• 2016

The purpose of this study is to identify the effects of a company's response strategies (response type, communication style, and response sincerity) on customer's brand attitude and purchase intentions. A fictional Facebook fan page containing 6 separate scenarios was developed based on actual customer reviews and company responses observed on Facebook restaurant fan pages. Participants were recruited from Amazon's Mechanical Turk (MTurk). A total of 202 responses were obtained; 185 responses were analyzed after deleting insufficient responses. The results of MANOVA found that an accommodative response leads customers to have a more favorable attitude towards a brand and have stronger purchasing intentions. In addition, customers who perceive the company's response to a negative review as sincere are more likely to have a positive brand attitude and purchasing intentions, as compared to those who perceive it as either insincere or neutral.

• International Journal of Contents
• /
• v.15 no.1
• /
• pp.1-9
• /
• 2019

Since the VR market is expected to have a high growth, this study aimed to investigate the human factor-related determinants of user satisfaction with mobile VR headsets. A pre-study of customer reviews was conducted with the help of semantic network analysis to identify the core keywords for understanding negative and positive predictors of mobile VR headset experiences. Through laboratory testing with three different commercial models, the main study measured and identified the predictors of user satisfaction. From the results, five factors were extracted as valid predictor variables and used for regression analysis. These factors were immersion, VR sickness, usability, wear-ability and menu navigation interface. All the five predictors were proved to be significant determinants of the perceived user satisfaction with mobile VR headsets. Usability was the strongest predictor, followed by VR sickness and wear-ability. Practical and theoretical implications of the results were discussed.

• International journal of advanced smart convergence
• /
• v.9 no.4
• /
• pp.124-131
• /
• 2020

As users of internet-based software applications increase, functional and non-functional problems for software applications are quickly exposed to user reviews. These user reviews are an important source of information for software improvement. User review mining has become an important topic of intelligent software engineering. This study proposes a user review mining method for software improvement. User review data collected by crawling on the app review page is analyzed to check user satisfaction. It analyzes the sentiment of positive and negative that users feel with a machine learning method. And it analyzes user requirement issues through topic analysis based on structural topic modeling. The user review mining process proposed in this study conducted a case study with the a non-face-to-face video conferencing app. Software improvement through user review mining contributes to the user lock-in effect and extending the life cycle of the software. The results of this study will contribute to providing insight on improvement not only for developers, but also for service operators and marketing.

• International Journal of Advanced Culture Technology
• /
• v.11 no.3
• /
• pp.190-198
• /
• 2023

Sentiment analysis can be applied to all texts generated from websites, blogs, messengers, etc. The study fulfills an artificial intelligence sentiment analysis estimating visiting evaluation opinions (reviews) and visitor ratings, and suggests a deep learning model which foretells either an affirmative or a negative inclination for new reviews. This study operates review big data about Jeju tourist attractions which are extracted from Google from October 1st, 2021 to November 30th, 2021. The normalization data used in the propensity prediction modeling of this study were divided into training data and test data at a 7.5:2.5 ratio, and the CNN classification neural network was used for learning. The predictive model of the research indicates an accuracy of approximately 84.72%, which shows that it can upgrade performance in the future as evaluating its error rate and learning precision.

• Information Systems Review
• /
• v.23 no.3
• /
• pp.127-158
• /
• 2021

Many online travel agencies (OTAs) provide average ratings and time-relevant information or the most recently posted reviews regarding hotels to satisfy customers. To identify these two factors' relative influence on behavioral decision-making processes, we conducted two studies: (1) an experimental research design to explore the relative influence of the two on online review consumption and (2) an empirical approach to examine their relative impact on online review generation. The results show that when review posters observe an inconsistency between average ratings and recent reviews, they tend to deviate from the recent reviews regardless of the overall direction (reactance behavior). Meanwhile, review consumers tend to conform to the opinions presented in recent reviews (herding behavior). Additionally, in both cases, the effects are amplified in case of a negative aberration. Based on the findings, this study provides theoretical and practical implications regarding the relative influences of average rating and recently posted reviews and their different impacts on online review consumption and generation.

• The Journal of the Korea Contents Association
• /
• v.12 no.4
• /
• pp.76-85
• /
• 2012

Through opinion mining, we can analyze the degree of positive or negative sentiments that customers feel about important entities or attributes in online customer reviews. But, the limit of the opinion mining techniques is to provide only simple functions in analyzing the reviews. In this paper, we proposed novel techniques that can analyze the online customer reviews multi-dimensionally. The novel technique is to modify the existing OLAP techniques so that they can be applied to text data. The novel technique, that is, multi-dimensional analytic model consists of noun, adjective and document axes which are converted into four relational tables in relational database. The multi-dimensional analysis model would be new framework which can converge the existing opinion mining, information summarization and clustering algorithms. In this paper, we implemented the multi-dimensional analysis model and algorithms. we recognized that the system would enable us to analyze the online customer reviews more complexly.