• Annual Conference of KIPS
• /
• 2020.05a
• /
• pp.389-392
• /
• 2020

The growth in social media, blogs and restaurant listing directories have led to increasing customer reviews about restaurants, their quality of food items and services available on the internet. These user reviews offer a massive amount of valuable information that can be used for various decision-making purposes. Currently, most food recommendation sites provide recommendation scores about restaurants rather than food items of the restaurant and the provided recommendation scores may be biased since they are calculated only from user reviews listed only in their sites. Usually, people wants a reliable recommendation about foods, not restaurant. In this paper, we present a reliable Korean food items rating method; we first extract food items by applying NER technique to restaurant reviews collected from many Korean restaurant recommendation web sites, blogs and web data. Then, we apply lexicon-based sentiment analysis on collected user reviews and predict people's opinions as sentiment polarity scores (+1 for positive; -1 for negative; 0 for neutral). Finally, by taking average of all calculated polarity scores about a food item, we obtain a rating to individual menu items of the restaurant. The proposed food item rating is more reliable since it does not depend on reviews of only one site.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.41 no.1
• /
• pp.1-10
• /
• 2018

Current evaluation practices for IT projects suffer from several problems, which include the difficulty of self-explanation for the evaluation results and the improperly scaled scoring system. This study aims to develop a methodology of opinion mining to extract key factors for the causal relationship analysis and to assess the feasibility of quantifying evaluation scores from text comments using opinion mining based on big data analysis. The research has been performed on the domain of publicly procured IT proposal evaluations, which are managed by the National Procurement Service. Around 10,000 sets of comments and evaluation scores have been gathered, most of which are in the form of digital data but some in paper documents. Thus, more refined form of text has been prepared using various tools. From them, keywords for factors and polarity indicators have been extracted, and experts on this domain have selected some of them as the key factors and indicators. Also, those keywords have been grouped into into dimensions. Causal relationship between keyword or dimension factors and evaluation scores were analyzed based on the two research models-a keyword-based model and a dimension-based model, using the correlation analysis and the regression analysis. The results show that keyword factors such as planning, strategy, technology and PM mostly affects the evaluation result and that the keywords are more appropriate forms of factors for causal relationship analysis than the dimensions. Also, it can be asserted from the analysis that evaluation scores can be composed or calculated from the unstructured text comments using opinion mining, when a comprehensive dictionary of polarity for Korean language can be provided. This study may contribute to the area of big data-based evaluation methodology and opinion mining for IT proposal evaluation, leading to a more reliable and effective IT proposal evaluation method.

• The Research Journal of the Costume Culture
• /
• v.28 no.2
• /
• pp.265-284
• /
• 2020

This study aimed to analyze the performance of Disney-collaborated fashion lines based on online consumer reviews. To do so, the researchers employed text mining and network analysis to identify key words in the reviews of these products. Blogs, internet cafes, and web documents provided by Naver, Daum, and YoutTube were selected as subjects for the analysis. The analysis period was limited to one year after for the 2019. Data collection and analysis were conducted using Python 3.7, Textom, and NodeXL. The research terms in question were as follows: 'Disney fashion collaboration' and 'Frozen fashion collaboration'. Preliminary survey results indicated that 'Elsa's dress' was the most frequently mentioned term and that the domestic fashion brand Eland Retail was the most active in selling Disney branded clothing through its own brand. The writers of reviews for Disney-collaborated fashion products were primarily mothers with daughters. Their decision to purchase these products was based upon the following factors; price, size, stability of decoration, shipping, laundry, and retailer. The motives for purchasing the product were the positive response of the consumer's child and the satisfaction of the parents due to the child's response. The problems to be solved included insufficient quantity of supply, delay in delivery, expensive price considering the number of times children's clothes are worn, poor glitter decoration, faded color, contamination from laundry, and undesirable smells immediately after the purchase.

• Journal of Korean Society for Quality Management
• /
• v.49 no.4
• /
• pp.595-607
• /
• 2021

Purpose: This study introduces a systematic framework to evaluate service satisfaction of cosmetic brands through online review analysis utilizing Text-Mining technique. Methods: The framework assumes that the service satisfaction is evaluated by positive comments from online reviews. That is, the service satisfaction of a cosmetic brand is evaluated higher as more positive opinions are commented in the online reviews. This study focuses on two approaches. First, it collects online review comments from the top 50 global cosmetic brands and evaluates customer service satisfaction for each cosmetic brands by applying Sentimental Analysis and Latent Dirichlet Allocation. Second, it analyzes the determinants that induce or influence service satisfaction and suggests the guidelines for cosmetic brands with low satisfaction to improve their service satisfaction. Results: For the satisfaction evaluation, online review data were extracted from the top 50 global cosmetic brands in the world based on 2018 sales announced by Brand Finance in the UK. As a result of the satisfaction analysis, it was found that overall there were more positive opinions than negative opinions and the averages for polarity, subjectivity, positive ratio, and negative ratio were calculated as 0.50, 0.76, 0.57, and 0.19, respectively. Polarity, subjectivity and positive ratio showed the opposite pattern to negative ratio, and although there was a slight difference in fluctuation range and ranking between them, the patterns are almost same. Conclusion: The usefulness of the proposed framework was verified through case study. Although some studies have suggested a method to analyze online reviews, they didn't deal with the satisfaction evaluation among competitors and cause analysis. This study is different from previous studies in that it evaluates service satisfaction from a relative point of view among cosmetic brands and analyze determinants.

• The Journal of the Korea Contents Association
• /
• v.21 no.10
• /
• pp.67-77
• /
• 2021

This study conducted topic modeling, association analysis, and sentiment analysis focused on text mining in order to reflect regional characteristics in the process of establishing an information plan in Chungcheongbuk-do. As a result of the analysis, it was confirmed that Chungcheongbuk-do occupies a relatively high proportion of educational activities to bridge the information gap, and is interested in improving infrastructure to provide non-face-to-face, untouched administrative services, and bridge the gap between urban and rural areas. In addition, it is necessary to refer to the fact that there is a positive evaluation of the combination of bio and IT in the regional strategic industry and examples of ICT innovation services. It has been confirmed that smart cities have high expectations for the establishment of various cooperation systems with IT companies, but continuous crisis management is necessary so that they are not related to political issues. It is hoped that the results of this study can be used as one of the methods to specifically reflect regional changes in the process of informatization.

• Journal of Computing Science and Engineering
• /
• v.6 no.2
• /
• pp.143-150
• /
• 2012

Computational techniques for topic classification can support qualitative research by automatically applying labels in preparation for qualitative analyses. This paper presents an evaluation of supervised learning techniques applied to one such use case, namely, that of labeling emotions, instructions and information in suicide notes. We train a collection of one-versus-all binary support vector machine classifiers, using cost-sensitive learning to deal with class imbalance. The features investigated range from a simple bag-of-words and n-grams over stems, to information drawn from syntactic dependency analysis and WordNet synonym sets. The experimental results are complemented by an analysis of systematic errors in both the output of our system and the gold-standard annotations.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.5
• /
• pp.47-56
• /
• 2020

The stock index is used not only as an economic indicator for a country, but also as an indicator for investment judgment, which is why research into predicting the stock index is ongoing. The task of predicting the stock price index involves technical, basic, and psychological factors, and it is also necessary to consider complex factors for prediction accuracy. Therefore, it is necessary to study the model for predicting the stock price index by selecting and reflecting technical and auxiliary factors that affect the fluctuation of the stock price according to the stock price. Most of the existing studies related to this are forecasting studies that use news information or macroeconomic indicators that create market fluctuations, or reflect only a few combinations of indicators. In this paper, this we propose to present an effective combination of the news information sentiment analysis and various macroeconomic indicators in order to predict the US Dow Jones Index. After Crawling more than 93,000 business news from the New York Times for two years, the sentiment results analyzed using the latest natural language processing techniques BERT and NLTK, along with five macroeconomic indicators, gold prices, oil prices, and five foreign exchange rates affecting the US economy Combination was applied to the prediction algorithm LSTM, which is known to be the most suitable for combining numeric and text information. As a result of experimenting with various combinations, the combination of DJI, NLTK, BERT, OIL, GOLD, and EURUSD in the DJI index prediction yielded the smallest MSE value.

• International Journal of Advanced Culture Technology
• /
• v.4 no.2
• /
• pp.7-18
• /
• 2016

The first D2C-era web-real variety show in Korea was broadcast via tvN of CJ E&M. The web-real variety program 'SinSeoYuGi' accumulated 54 million views, along with 50 million views at the Chinese portal site QQ. This study carries out an analysis using text mining that extracts portal site blogs, twitter page views and associative terms. In addition, this study derives viewers' response by extracting key words with opinion mining techniques that divide positive words, neutral words and negative words through customer sentiment analysis. It is found that the success factors of the web-real variety were reduced in appearance fees and production cost, harmony between actual cast members and scenario characters, mobile TV programing, and pre-roll advertising. It is expected that web-real variety broadcasting will increase in value as web contents in the future, and be established as a new genre with the job of 'technical marketer' growing as well.

• Human Ecology Research
• /
• v.60 no.1
• /
• pp.21-38
• /
• 2022

This study uses text mining compares and contrasts consumers' social media discourses on dietary related zero-waste movement before and after COVID-19. The results indicate that the amount of buzz on social networks for the zero- waste movement has been increasing after COVID-19. Additionally, the results of frequency analysis and topic modeling revealed that subjects associated with zero-waste movement were more diversified after COVID-19. Although the results of a sentiment analysis and word cloud visualization confirmed that consumers' positive responses toward the zero-waste have been increasing, they also revealed a need to educate and encourage those who are still not aware of the need for zero-waste. Finally, consumers mentioned only a small number of companies participating in zero-waste movement on SNS, indicating that the level of active involvement by such companies is much lower than that of consumers. Theoretical and educational implications as well as those for government policy-making are considered.

• International Journal of Computer Science & Network Security
• /
• v.23 no.11
• /
• pp.21-31
• /
• 2023

In this paper, we focused on the problem of evaluating multi-class classification accuracy and simulation of multiple classifier performance metrics. Multi-class classifiers for sentiment analysis involved many challenges, whereas previous research narrowed to the binary classification model since it provides higher accuracy when dealing with text data. Thus, we take inspiration from the non-linear Support Vector Machine to modify the algorithm by embedding dynamic hyperplanes representing multiple class labels. Then we analyzed the performance of multi-class classifiers using macro-accuracy, micro-accuracy and several other metrics to justify the significance of our algorithm enhancement. Furthermore, we hybridized Enhanced Convolution Neural Network (ECNN) with Dynamic Support Vector Machine (DSVM) to demonstrate the effectiveness and efficiency of the classifier towards multi-class text data. We performed experiments on three hybrid classifiers, which are ECNN with Binary SVM (ECNN-BSVM), and ECNN with linear Multi-Class SVM (ECNN-MCSVM) and our proposed algorithm (ECNNDSVM). Comparative experiments of hybrid algorithms yielded 85.12 % for single metric accuracy; 86.95 % for multiple metrics on average. As for our modified algorithm of the ECNN-DSVM classifier, we reached 98.29 % micro-accuracy results with an f-score value of 98 % at most. For the future direction of this research, we are aiming for hyperplane optimization analysis.