• Annual Conference on Human and Language Technology
• /
• 2021.10a
• /
• pp.379-383
• /
• 2021

감정 분석은 텍스트 데이터에서 인간이 느끼는 감정을 다양한 감정 유형으로 분류하는 것이다. 그러나 많은 연구에서 감정 분석은 긍정과 부정, 또는 중립의 극성을 분류하는 감성 분석의 개념과 혼용되고 있다. 본 연구에서는 텍스트에서 느껴지는 감정들을 다양한 감정 유형으로 분류한 감정 말뭉치를 구축하였는데, 감정 말뭉치를 구축하기 위해 심리학 모델을 기반으로 분류한 감정 어휘 사전을 사용하였다. 9가지 감정 유형으로 분류된 한국어 감정 어휘 사전을 바탕으로 한국어 영화 리뷰 말뭉치에 9가지 감정 유형의 감정을 태깅하여 감정 분석 말뭉치를 구축하고, KcBert에 학습시켰다. 긍정과 부정으로 분류된 데이터로 사전 학습된 KcBert에 9개의 유형으로 분류된 데이터를 학습시켜 기존 모델과 성능 비교를 한 결과, KcBert는 다중 분류 모델에서도 우수한 성능을 보였다.

• Annual Conference on Human and Language Technology
• /
• 2019.10a
• /
• pp.519-521
• /
• 2019

최근 소셜 메신저를 통해 많은 사람들이 의사소통을 주고받음에 따라, 텍스트에서 감정을 파악하는 것이 중요하다. 따라서, 감정이 태깅된 데이터가 필요하다. 하지만, 기존 연구는 감정이 태깅된 데이터의 양이 많지가 않다. 이는 텍스트에서 감정을 파악하는데 성능 저하를 야기할 수 있다. 이를 해결하기 위해, 본 논문에서는 단어 매칭 방법과 형태소 매칭 방법을 이용하여 많은 양의 한국어 감정 태깅 데이터셋인 EmoNSMC 를 구축하였다. 구축한 데이터셋은 네이버 영화 감상 리뷰 데이터 (NSMC)에 디스턴트 수퍼비전 방법 (distant supervision) 방법을 적용하여 weak labeling을 진행하였고, 이 과정에서 한국어 감정 어휘 사전 (KTEA) 을 이용하였다. 구축된 데이터셋의 감정 분포 결과, 형태소 매칭 방법을 통해 구축한 데이터셋이 좀 더 감정 분포가 균등한 것을 확인할 수 있었다. 해당 데이터셋은 공개되어 있다.

• Annual Conference on Human and Language Technology
• /
• 2022.10a
• /
• pp.576-581
• /
• 2022

감정 분석은 긍부정의 극성을 판단하는 감성 분석과 달리 텍스트로부터 구체적인 감정 유형을 분류해내는 과제이다. 본 논문에서는 소설 텍스트에 감정 분석을 수행하는 것을 새로운 과제로 설정하고, 이에 활용할 수 있는 감정 용언 사전을 소개한다. 이 사전에는 맥락과 상관없이 동일한 감정을 전달하는 직접 감정 표현과 맥락에 따라 다른 감정으로 해석될 수 있는 간접 감정 표현이 구분되어 있다. 우리는 이로써 한국어 자연어처리 연구자들이 소설의 풍부한 감정 표현 텍스트로부터 정확한 감정을 분류해낼 수 있도록 그 단초를 마련한다.

• 한국HCI학회:학술대회논문집
• /
• 2007.02a
• /
• pp.938-942
• /
• 2007

신문 기사 및 상품 평은 특정 주제나 상품을 대상으로 하여 글쓴이의 감정과 의견이 잘 나타나 있는 대표적인 문서이다. 최근 여론 조사 및 상품 의견 조사 등 다양한 측면에서 대용량의 문서의 의미적 분류 및 분석이 요구되고 있다. 본 논문에서는 문서에 나타난 내용을 기준으로 문서가 나타내고 있는 감정을 긍정과 부정의 두 가지 범주로 분류하는 시스템을 구현한다. 문서 분류의 시작은 감정을 지닌 대표적인 종자 어휘(seed word)로부터 시작하며, 자질의 선정은 한국어 특징상 감정 및 감각을 표현하는 명사, 형용사, 부사, 동사를 대상으로 한다. 가중치 부여 방법은 한글 유의어 사전을 통해 종자 어휘의 의미를 확장하여 각각의 가중치를 책정한다. 단어 벡터로 표현된 입력 문서를 이진 분류기인 지지벡터 기계를 이용하여 문서에 나타난 감정을 판단하는 시스템을 구현하고 그 성능을 평가한다.

• The Journal of Society for e-Business Studies
• /
• v.19 no.4
• /
• pp.1-19
• /
• 2014

Efforts to identify user's recognition which exists in the big data are being conducted actively. They try to measure scores of people's view about products, movies and social issues by analyzing statements raised on Internet bulletin boards or SNS. So this study deals with the problem of determining how to find the emotional vocabulary and the degree of these values. The survey methods are using the results of previous studies for the basic emotional vocabulary and degree, and inferring from the dictionary's glosses for the extended emotional vocabulary. The results were found to have the 4 emotional words lists (vocabularies) as basic emotional list, extended 1 stratum 1 level list from basic vocabulary's glosses, extended 2 stratum 1 level list from glosses of non-emotional words, and extended 2 stratum 2 level list from glosses' glosses. And we obtained the emotional degrees by applying the weight of the sentences and the emphasis multiplier values on the basis of basic emotional list. Experimental results have been identified as AND and OR sentence having a weight of average degree of included words. And MULTIPLY sentence having 1.2 to 1.5 weight depending on the type of adverb. It is also assumed that NOT sentence having a certain degree by reducing and reversing the original word's emotional degree. It is also considered that emphasis multiplier values have 2 for 1 stratum and 3 for 2 stratum.

• Journal of KIISE:Computing Practices and Letters
• /
• v.14 no.3
• /
• pp.336-340
• /
• 2008

Sentiment classification is a recent subdiscipline of text classification, which is concerned not with the topic but with opinion. In this paper, we present a Korean sentence and document classification system using effective sentiment features. Korean sentiment classification starts from constructing effective sentiment feature sets for positive and negative. The synonym information of a English word thesaurus is used to extract effective sentiment features and then the extracted English sentiment features are translated in Korean features by English-Korean dictionary. A sentence or a document is represented by using the extracted sentiment features and is classified and evaluated by SVM(Support Vector Machine).

• Smart Media Journal
• /
• v.9 no.4
• /
• pp.144-151
• /
• 2020

In a mobile environment, communication takes place via SMS text messages. Vocabularies used in SMS texts can be expected to use vocabularies of different classes from those used in general Korean literary style sentence. For example, in the case of a typical literary style, the sentence is correctly initiated or terminated and the sentence is well constructed, while SMS text corpus often replaces the component with an omission and a brief representation. To analyze these vocabulary usage characteristics, the existing colloquial style corpus and the literary style corpus are used. The experiment compares and analyzes the vocabulary use characteristics of the colloquial corpus SMS text corpus and the Naver Sentiment Movie Corpus, and the written Korean written corpus. For the comparison and analysis of vocabulary for each corpus, the part of speech tag adjective (VA) was used as a standard, and a distinctive collexeme analysis method was used to measure collostructural strength. As a result, it was confirmed that adjectives related to emotional expression such as'good-','sorry-', and'joy-' were preferred in the SMS text corpus, while adjectives related to evaluation expressions were preferred in the Naver Sentiment Movie Corpus. The word embedding was used to automatically construct a sentiment lexicon based on the extracted adjectives with high collostructural strength, and a total of 343,603 sentiment representations were automatically built.

• Annual Conference on Human and Language Technology
• /
• 2022.10a
• /
• pp.238-244
• /
• 2022

감정 분석은 텍스트에 표현된 인간의 감정을 인식하여 다양한 감정 유형으로 분류하는 것이다. 섬세한 인간의 감정을 보다 정확히 분류하기 위해서는 감정 유형의 분류가 무엇보다 중요하다. 본 연구에서는 사전 학습 언어 모델을 활용하여 우리말샘의 감정 어휘와 용례를 바탕으로 기쁨, 슬픔, 공포, 분노, 혐오, 놀람, 흥미, 지루함, 통증의 감정 유형으로 분류된 감정 말뭉치를 구축하였다. 감정 말뭉치를 구축한 후 성능 평가를 위해 대표적인 트랜스포머 기반 사전 학습 모델 중 RoBERTa, MultiDistilBert, MultiBert, KcBert, KcELECTRA. KoELECTRA를 활용하여 보다 넓은 범위에서 객관적으로 모델 간의 성능을 평가하고 각 감정 유형별 정확도를 바탕으로 감정 유형의 특성을 알아보았다. 그 결과 각 모델의 학습 구조가 다중 분류 말뭉치에 어떤 영향을 주는지 구체적으로 파악할 수 있었으며, ELECTRA가 상대적으로 우수한 성능을 보여주고 있음을 확인하였다. 또한 감정 유형별 성능을 비교를 통해 다양한 감정 유형 중 기쁨, 슬픔, 공포에 대한 성능이 우수하다는 것을 알 수 있었다.

• Korean Journal of Cognitive Science
• /
• v.19 no.4
• /
• pp.499-517
• /
• 2008

In this paper, we propose an effective emotion feature extraction method for Korean and evaluate their availability in sentiment classification. Korean emotion features are expanded from several representative emotion words and they play an important role in building in an effective sentiment classification system. Firstly, synonym information of English word thesaurus is used to extract effective emotion features and then the extracted English emotion features are translated into Korean. To evaluate the extracted Korean emotion features, we represent each document using the extracted features and classify it using SVM(Support Vector Machine). In experimental results, the sentiment classification system using the extracted Korean emotion features obtained more improved performance(14.1%) than the system using content-words based features which have generally used in common text classification systems.

• Cross-Cultural Studies
• /
• v.53
• /
• pp.131-161
• /
• 2018

Emotional expressions are expressions that show the internal condition of mind or consciousness. Types of emotional expressions include vocabulary that describes emotion, the composition of sentences that expresses emotion such as an exclamatory sentence and rhetorical question, expressions of interjection, appellation, causative, passive, adverbs of attitude for an idea, and a style of writing. This study focuses on vocabulary that describes emotion and analyzes the aspect of translation when emotional expressions of 'Hu(怖)' is shown on "Kokoro". The aspect of translation was analyzed by three categories as follows; a part of speech, handling of subjects, and classification of meanings. As a result, the aspect of translation for expressions of Hu(怖)' showed that they were translated to vocabulary as they were suggested in the dictionary in some cases. However, they were not always translated as they were suggested in the dictionary. Vocabulary that described the emotion of 'Hu(怖)' in Japanese sentences were mostly translated to their corresponding parts of speech in Korean. Some adverbs needed to add 'verbs' when they were translated. Also, different vocabulary was added or used to maximize emotion. However, the correspondence of a part of speech in English was different from Korean. Examples of Japanese sentences that expressed 'Hu(怖)' by verbs were translated to expression of participles for passive verbs such as 'fear', 'dread', 'worry', and 'terrify' in many cases. Also, idioms were translated with focus on the function of sentences rather than the form of sentences. Examples, what was expressed in adverbs did not accompany verbs of 'Hu (怖)'. Instead, it was translated to the expression of participles for passive verbs and adjectives such as 'dread', 'worry', and 'terrify' in many cases. The main agents of emotion were shown in the first person and the third person in simple sentences. The translation on emotional expressions when a main agent was the first person showed that the fundamental word order of Japanese was translated as it was in Korean. However, adverbs of time and adverbs of degree tended to be added. Also, the first person as the main agent of emotion was positioned at the place of subject when it was translated in English. However, things or the cause of events were positioned at the place of subject in some cases to show the degree of 'Hu(怖)' which the main agent experienced. The expression of conjecture and supposition or a certain visual and auditory basis was added to translate the expression of emotion when the main agent of emotion was the third person. Simple sentences without a main agent of emotion showed that their subjects could be omitted even if they were essential components because they could be known through context in Korean. These omitted subjects were found and translated in English. Those subjects were not necessarily humans who were the main agents of emotion. They could be things or causes of events that specified the expression of emotion.