• Annual Conference on Human and Language Technology
• /
• 2020.10a
• /
• pp.507-511
• /
• 2020

오픈 도메인 대화에서 텍스트에 나타난 태도나 성향과 같은 화자의 주관적인 감정정보를 분석하는 것은 사용자들에게서 풍부한 응답을 이끌어 내고 동시에 제공하는 목적으로 사용될 수 있다. 하지만 한국어 감성분석에서 기존의 대부분의 연구들은 긍정과 부정 두개의 클래스 분류만을 다루고 있고 이는 현실 화자의 감정 정보를 정확하게 분석하기에는 어려움이 있다. 또한 최근에 오픈한 다중클래스로된 한국어 대화 감성분석 데이터셋은 중립 클래스가 전체 데이터셋의 절반을 차지하고 일부 클래스는 사용하기에 매우 적은, 다시 말해 클래스 간의 데이터 불균형 문제가 있어 다루기 굉장히 까다롭다. 이 논문에서 우리는 일곱개의 클래스가 존재하는 한국어 대화에서 세션들을 효율적으로 분류하는 기법들에 대해 논의한다. 우리는 극심한 클래스 불균형에도 불구하고 76.56 micro F1을 기록하였다.

• Science of Emotion and Sensibility
• /
• v.11 no.4
• /
• pp.589-598
• /
• 2008

The purpose of this study was to compare the emotional responses of people with visual disability with those of normal people and to understand their similarity or differences in order to apply the new understandings into the future research on Web Accessibility Guidelines. For this purpose, a Web survey system was developed using 15 auditorial stimuli prepared based on the Media Taxonomy and 11 emotion measuring criteria selected from the literature review. After developing the system, emotional responses of 31 people with visual disability and 53 normal people were collected through the Web. The results of the survey showed that the emotional responses of people with visual disability were similar to those of normal people, although there were some exceptional cases. Therefore, it is clear that emotional needs of people with disability should be taken count of in the Web accessibility discussions and further in-depth studies on the emotional characteristics of people with disability are necessary.

• Journal of Korea Society of Industrial Information Systems
• /
• v.27 no.1
• /
• pp.49-62
• /
• 2022

Online customer review data can be easily collected on the Internet and also they describe sentimental evaluation of a product in different aspects. Previous sentiment analysis studies evaluate the degree of sentiment with review data, which may have multiple sentences describing different product aspects. Since different aspects of a product can be described in a sentence, the proposed method suggested analyzing a sentence to build a pair of a product aspect terms and sentimental terms. Bidirectional LSTM and CRF algorithms were used in this paper. A pair of aspect terms and sentimental terms are evaluated by pre-defined evaluation rules. The paper suggested using the result of evaulation as inputs of QFD, so that the quantified customer voices effect on the requirements of a new product. Online reviews for a hair dryer were used as an example showing that the proposed approach can derive reasonable sentiment analysis results.

• The Journal of the Korea Contents Association
• /
• v.16 no.2
• /
• pp.332-343
• /
• 2016

With the advance of digital technology, new methods which acquire color information and combine it with various contents are emerging. Color has some effect on emotion while it gives some information as component of an image. In addition, change in emotion and sensation from color stimulus makes some change in cognition. This research investigate the effect of color overlay on cognition depending on emotional state. With this goal, subjects consisting of 10 men and 10 women solved some problems with color overlay of red, orange, and green after watching short video clips which intend to induce target emotion. Experimental results show that red color overlay under positive emotion significantly reduces the average score of solving problems, while green overlay under negative emotion significantly increases it. It is also analyzed that there is not statistically significant difference in cognitive function with color overlay while it is significantly better under positive emotion than negative emotion without color overlay.

• Journal of the Korea Convergence Society
• /
• v.8 no.4
• /
• pp.1-8
• /
• 2017

In the modern society, the size of the fashion market is continuously increasing both overseas and domestic. When purchasing a product through e-commerce, the evaluation data for the product created by other consumers has an effect on the consumer's decision to purchase the product. By analysing the consumer's evaluation data on the product the company can reflect consumer's opinion which can leads to positive affect of performance to company. In this paper, we propose a method to construct a model to analyze user's sentiment using word embedding space formed by learning review data of amazon fashion products. Experiments were conducted by learning three SVM classifiers according to the number of positive and negative review data using the formed word embedding space which is formed by learning 5.7 million Amazon review data.. Experimental results showed the highest accuracy of 88.0% when learning SVM classifier using 50,000 positive review data and 50,000 negative review data.

• Science of Emotion and Sensibility
• /
• v.13 no.1
• /
• pp.243-250
• /
• 2010

A study is presented that examines the effect of personality on preference and purchase intention of positive and negative novelty design products. Participants evaluated their preference and purchase intention on 18 design products including positive and negative novelty design products. A cute Scream character was created as a positive novelty design and a skeleton Mickey character was used as a negative one. We measured extroversion, neuroticism, and visual product aesthetics as individual traits. The results shows that people, of course, prefer(and want to buy) the positive novelty design product to the negative novelty design product. Only in the case of the negative novelty design product, the more neurotic people is, the more they prefer it. Also, among the people who like the negative novelty design product, introvert ones are likely to purchase it but extroverted ones are not. These findings imply that personality plays an important role in liking and purchasing negative novelty design product.

• Science of Emotion and Sensibility
• /
• v.9 no.4
• /
• pp.321-329
• /
• 2006

Recently there has been interested in the role of emotion in human computer interaction fields. The present study investigated whether the users emotional state effects on information search pattern for decision making in small screen display. In experiment, to induce specific emotional states(positive and negative emotional state), the participants were asked to listen to music and imagine autobiographic events with different emotional impacts. Subsequently, they performed time limit search tasks with three travel information on small screen display and their search patterns were recorded on real time. The results indicated that a positive emotional state caused more wide and fast Information search pattern in comparison with neutral and negative emotional state. And neutral and negative emotional state caused more cognitive resource to details in comparison with positive emotional state.

• Culinary science and hospitality research
• /
• v.19 no.3
• /
• pp.173-193
• /
• 2013

This study was conducted to investigate the effects of servicescape factors(space, convenience, aesthetics, comfortableness, cleanliness), brand reputation and experiences factors(feel, think, act, relate) on emotional and cognitive reactions of visitors to Korean restaurants. It also investigated the relationship between emotional and cognitive reactions and brand loyalty. A total of 353 questionnaires were analyzed using the factor analysis, a reliability test, and structural equation modeling analysis. The results revealed that space, aesthetics and comfortableness were positively related to emotional reaction, and convenience and cleanliness were related to cognitive reaction. This study also found that brand reputation was positively related to emotional and cognitive reactions. Moreover, it was found that think was positively related to emotional reaction, and feel, think, act and relate were positively related to cognitive reaction. Finally, cognitive reaction was positively related to loyalty. Therefore, Korean restaurant marketers should introduce or strengthen servicescape, brand reputation and experiences, which increases brand attitude or loyalty.

• The Journal of Korean Association of Computer Education
• /
• v.5 no.3
• /
• pp.107-116
• /
• 2002

In order to enhance instructional effects of multimedia contents, it is critical to emphasize the importance of interface and the affective aspects of learner characteristics in design processes. The main aim of this research was to assess the needs for evaluating icon design strategies on the basis of children's sensibilities by investigating differences between children and teachers in their sensibilities to the icons used in multimedia contents for elementary education. Some interesting findings were obtained from the research. Children had richer sensibilities to the icons presented in multimedia contents than teachers in terms of the average number of adjectives reflecting sensibilities. Among the children group, children living in urban areas showed richer sensibilities to the contents than those in rural areas. When the frequencies of the use of 5 adjectives commonly used by the two groups were compared, teachers showed positive sensibilities, but children demonstrated some negative sensibilities. Also children demonstrated richer sensibilities to certain contents than teachers. The results of the research indicated the need for evaluating icon design strategies on the basis of children's sensibilities in designing multimedia contents for children.

• Journal of Advanced Marine Engineering and Technology
• /
• v.39 no.3
• /
• pp.274-279
• /
• 2015

The main goal of this research is to build a sentiment analysis system which automatically determines user opinions of the Stanford Sentiment Treebank in terms of three sentiments such as positive, negative, and neutral. Firstly, sentiment sentences are POS tagged and parsed to dependency structures. All nodes of the Treebank and their polarities are automatically extracted from the Treebank. We train two Support Vector Machines models. One is for a node level classification and the other is for a sentence level. We have tried various type of features such as word lexicons, POS tags, Sentiment lexicons, head-modifier relations, and sibling relations. Though we acquired 74.2% in accuracy on the test set for 3 class node level classification and 67.0% for 3 class sentence level classification, our experimental results for 2 class classification are comparable to those of the state of art system using the same corpus.