• Proceedings of the Korean Society for Emotion and Sensibility Conference
• /
• 2007.05a
• /
• pp.109-113
• /
• 2007

본 논문은 칼라 패턴의 감성 평가를 위해 러프 집합 이론이 효과적으로 사용될 수 있음을 보여준다. 우리는 주어진 랜덤 칼라 패턴을 보여주고 사람들로 하여금 감성 평가를 하게 하여 수집된 심리학적 실험 데이터를 기반으로 VPRS(Variable Precision Rough Set) 이론을 적용, 관련 규칙들을 추출하였다. 이러한 규칙들은 벽지 등의 컬러 패턴들에 대한 근사적인 감성 평가 뿐만 아니라, 이미지 속성 공간을 언어적 이미지 스케일로 표현된 감성 공간으로 매핑 시키기 위한 적응 퍼지 시스템 등의 초기 조건으로도 사용할 수도 있다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2014.10a
• /
• pp.438-440
• /
• 2014

Recently it is used commercially to actively move the data from the SNS service. Therefore, we propose a method that can accurately analyze the information related to the reputation of companies and products in real time SNS environment in this paper.Identify the relationship between words by performing morphological analysis on the text data gathered by crawling the SNS scheme. In addition, it shows the visualization to analyze statistically through a established emotional dictionary morphemes are extracted from the sentence. Here, if the extracted word is not exist in sentimental dictionary. Also, we propose the algorithm that add the word to emotional dictionary automatically.

• 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.

• Journal of Internet Computing and Services
• /
• v.24 no.4
• /
• pp.127-135
• /
• 2023

This research paper explores the application of FinBERT, a variational BERT-based model pre-trained on financial domain, for sentiment analysis in the financial domain while focusing on the process of identifying suitable training data and hyperparameters. Our goal is to offer a comprehensive guide on effectively utilizing the FinBERT model for accurate sentiment analysis by employing various datasets and fine-tuning hyperparameters. We outline the architecture and workflow of the proposed approach for fine-tuning the FinBERT model in this study, emphasizing the performance of various datasets and hyperparameters for sentiment analysis tasks. Additionally, we verify the reliability of GPT-3 as a suitable annotator by using it for sentiment labeling tasks. Our results show that the fine-tuned FinBERT model excels across a range of datasets and that the optimal combination is a learning rate of 5e-5 and a batch size of 64, which perform consistently well across all datasets. Furthermore, based on the significant performance improvement of the FinBERT model with our Twitter data in general domain compared to our news data in general domain, we also express uncertainty about the model being further pre-trained only on financial news data. We simplify the complex process of determining the optimal approach to the FinBERT model and provide guidelines for selecting additional training datasets and hyperparameters within the fine-tuning process of financial sentiment analysis models.

• Archives of design research
• /
• v.17 no.4
• /
• pp.141-148
• /
• 2004

The purpose of this study is to define the emotion that expressed in step animation and to quantify the intuitional expression of emotion that related step for using extract, measure, analysis the stimulate element about step. The survey of relation with 27 word of emotional expressions and 36 moving pictures of step sample is used for method of this test. The emotional mental structure is transferred to 2 dimensional planes as applying the results of analysis of integrated data using Quantification Method 3, which the integrated data is composed two axial - confidential axial and stabling axial. Analysis of distribution of 2 dimensional diagram shows that the second of the plane and the third of the plane have much data. However, the first of the plane and the forth of the plane have a little data. Through this kind of analysis of graph, it is difficult to express a different emotion between unstable the timidity mind and stable feel the timidity mind using only step analysis. Six difference types about physical elements affecting to emotion are selected and analyzed such as the paces of step, the rate of step, the movement angle of pelvis, the swing range of arm, angle of backbone and the lean angle of body. The result is that the rate of stop and the lean angle of body are the major element that effects to emotional stimulate of stop. This thesis argues about methods transforming subjective expression to objective and quantitative expression with the state of delicate emotion of character apply to step animation naturally. Those data to apply to multi-contents in future are the main target in this study.

• Journal of KIISE
• /
• v.44 no.11
• /
• pp.1149-1155
• /
• 2017

Recently, with the development of communication technology, it has become possible to collect various sensor data that indicate the environmental stimuli within a space. In this paper, we propose a group emotion prediction system using a modular Bayesian network that was designed considering the psychological impact of environmental stimuli. A Bayesian network can compensate for the uncertain and incomplete characteristics of the sensor data by the probabilistic consideration of the evidence for reasoning. Also, modularizing the Bayesian network has enabled flexible response and efficient reasoning of environmental stimulus fluctuations within the space. To verify the performance of the system, we predict public emotion based on the brightness, volume, temperature, humidity, color temperature, sound, smell, and group emotion data collected in a kindergarten. Experimental results show that the accuracy of the proposed method is 85% greater than that of other classification methods. Using quantitative and qualitative analyses, we explore the possibilities and limitations of probabilistic methodology for predicting group emotion.

• Annual Conference on Human and Language Technology
• /
• 2019.10a
• /
• pp.429-433
• /
• 2019

스마트폰 보급의 확산으로 제품 구매 시 웹 사이트 및 SNS를 이용하여 제품 리뷰를 참고하는 소비자들이 증가하고 있다. 전자 상거래 사이트의 제품 리뷰는 구매 예정자들에게 유용한 정보로 활용되곤 한다. 하지만 구매 예정자가 직접 제품에 대한 리뷰 데이터를 찾아 전체 내용을 일일이 읽고 분석해야하기 때문에 시간이 오래 걸릴뿐만 아니라 가공되지 않는 데이터가 줄 수 있는 정보는 한정적이다. 또한 이러한 리뷰들은 상품의 특징을 파악하기에도 어려움이 있다. 본 논문에서는 제품의 주요 이슈를 추출하고 주요 이슈에 대한 감성 분석과 감성 요약을 통해 제품 분석 및 평가를 제공하는 시스템을 설계 및 구현하였다. 이를 휴대폰 제품에 적용하여 구축한 시스템을 통해 소비자가 방대한 양의 제품의 리뷰 데이터를 분석할 필요 없이 제품의 주요 이슈와 가공된 분석 결과를 시각적으로 빠르게 제공받을 수 있음을 보였다.

• Annual Conference on Human and Language Technology
• /
• 2013.10a
• /
• pp.33-38
• /
• 2013

소셜미디어를 통한 여론분석과 브랜드 모니터링에 대한 요구가 증가하면서, 빅데이터로부터 감성을 분석하는 기술에 대한 필요가 늘고 있다. 이를 위해, 본 논문에서는 단순 긍/부정 감성이 아닌 20종류의 세분화된 감성을 분석하기 위한 감성어휘 구축 알고리즘을 제시한다. 감성어휘 구축을 위해서는 준지도학습을 사용하였으며, 도메인에 특화되지 않은 일반 감성어휘를 구축하도록 학습되었다. 학습된 감성어휘를 인물, 스마트기기, 정책 등 다양한 도메인의 트위터 데이터에 적용하여 세부감성을 분석한 결과, 알고리즘의 특성상 재현율이 낮다는 한계를 가지고 있었으나, 대부분의 감성에 대해 높은 정확도를 지닌 감성어휘를 구축할 수 있었고, 감성을 직간접적으로 나타내는 표현들을 학습할 수 있었다.

• Journal of Digital Convergence
• /
• v.17 no.9
• /
• pp.47-54
• /
• 2019

This study confirms the polarity of news articles on apartment prices using Opinion Mining which has widely been used for a big data analysis. The analyses were carried out utilizing internet news articles posted on the Naver for two years: 2012 and 2018. We proposed a sentiment analysis model and modeled a topic-oriented sentiment dictionary construction methods. As a result of analyzing the proposed sentiment analysis model, it was confirmed that there was a difference according to the tendency of the media companies in selecting social issues at the time of rising apartment prices. At the same time, we were able to find more affirmative articles in the media companies which share similar sentiment with the government in charge. In this paper, we proposed a sentiment analysis model that can be used in real estate field and analyzed the polarity of unformatted data related to real estate. In order to integrate them into various fields in the future, it is necessary to build the sentiment dictionaries by themes, as well as to collect various unformatted data over extended periods.

• Science of Emotion and Sensibility
• /
• v.12 no.2
• /
• pp.205-214
• /
• 2009

This paper addresses analyzing affective color data for emotional interior design. Both the physical and psychological patterns for spatial colors were tested on thirty subjects, of which fifteen were male. All subjects participated in both the physiological and psychological experiments. The data on the reflecting subjects' affective moods is gathered through EEG physical experiments and SD (Semantic Differential Scale) method surveys. This research has suggested the relation of both experiments through affective color response. The methods of SPSS 10.0 and TeleScan Version 2 are used for analyzing response data to coordinate the colour palette with changeable moods. From the analysis of statistical data, all of the visual stimuli related emotional keywords and physiological responses. Finally, the initial goal of this research is to construct an affective colour database that is tested through human color perception by physical and psychological experiments.