• 산업경영시스템학회지
• /
• 제46권1호
• /
• pp.32-41
• /
• 2023

Recently, many studies are being conducted to extract emotion from text and verify its information power in the field of finance, along with the recent development of big data analysis technology. A number of prior studies use pre-defined sentiment dictionaries or machine learning methods to extract sentiment from the financial documents. However, both methods have the disadvantage of being labor-intensive and subjective because it requires a manual sentiment learning process. In this study, we developed a financial sentiment dictionary that automatically extracts sentiment from the body text of analyst reports by using modified Bayes rule and verified the performance of the model through a binary classification model which predicts actual stock price movements. As a result of the prediction, it was found that the proposed financial dictionary from this research has about 4% better predictive power for actual stock price movements than the representative Loughran and McDonald's (2011) financial dictionary. The sentiment extraction method proposed in this study enables efficient and objective judgment because it automatically learns the sentiment of words using both the change in target price and the cumulative abnormal returns. In addition, the dictionary can be easily updated by re-calculating conditional probabilities. The results of this study are expected to be readily expandable and applicable not only to analyst reports, but also to financial field texts such as performance reports, IR reports, press articles, and social media.

• 산업경영시스템학회지
• /
• 제46권4호
• /
• pp.142-151
• /
• 2023

Sentiment analysis is a method used to comprehend feelings, opinions, and attitudes in text, and it is essential for evaluating consumer feedback and social media posts. However, creating sentiment dictionaries, which are necessary for this analysis, is complex and time-consuming because people express their emotions differently depending on the context and domain. In this study, we propose a new method for simplifying this procedure. We utilize syntax analysis of the Korean language to identify and extract sentiment words based on the Reason-Sentiment Pattern, which distinguishes between words expressing feelings and words explaining why those feelings are expressed, making it applicable in various contexts and domains. We also define sentiment words as those with clear polarity, even when used independently and exclude words whose polarity varies with context and domain. This approach enables the extraction of explicit sentiment expressions, enhancing the accuracy of sentiment analysis at the attribute level. Our methodology, validated using Korean cosmetics review datasets from Korean online shopping malls, demonstrates how a sentiment dictionary focused solely on clear polarity words can provide valuable insights for product planners. Understanding the polarity and reasons behind specific attributes enables improvement of product weaknesses and emphasis on strengths. This approach not only reduces dependency on extensive sentiment dictionaries but also offers high accuracy and applicability across various domains.

• 한국정보과학회논문지:소프트웨어및응용
• /
• 제37권12호
• /
• pp.946-950
• /
• 2010

기계 학습 기반의 많은 감정 분류 시스템들은 문장으로부터 언어적 자질을 추출하기 위하여 형태소 분석기를 사용한다. 그러나 온라인 상품평에는 많은 띄어쓰기 오류 및 철자 오류가 포함되어 있어서 일반적으로 형태소 분석기가 좋은 성능을 내기 어려우며, 기반 시스템의 낮은 성능은 감정 분류 시스템의 성능하락을 초래한다. 이러한 문제를 해결하기 위하여 본 논문에서는 어절 패턴과 음운 패턴의 최장 일치 매칭(matching)에 기반한 자질 추출 방법을 제안한다. 두 종류의 패턴은 대용량의 품사 부착 말뭉치로부터 자동으로 구축된다. 어절 패턴은 영사, 동사와 같은 내용어를 포함하는 어절들로 구성되며, 음운 패턴은 동사나 형용사와 같은 용언의 초성과 중성의 쌍으로 구성된다. 음운 패턴에 초성과 중성만을 사용한 이유는 철자 오류에 영향을 덜 받기 때문이다. 제안 방법을 평가하기 위하여 SVM(Support Vector Machine)을 기계 학습기로 사용하는 감정 분류 시스템을 구현하였다. 한국어 상품평에 대한 실험에서 제안 방법을 자질 추출 모듈로 사용하는 감정 분류 시스템이 형태소 분석기를 사용하는 것보다 우수한 성능을 보였다.

• ETRI Journal
• /
• 제38권3호
• /
• pp.494-501
• /
• 2016

Sentiment analysis incorporates natural language processing and artificial intelligence and has evolved as an important research area. Sentiment analysis on product reviews has been used in widespread applications to improve customer retention and business processes. In this paper, we propose a method for performing an intensified sentiment analysis on customer product reviews. The method involves the extraction of two feature sets from each of the given customer product reviews, a set of acoustic features (representing emotions) and a set of lexical features (representing sentiments). These sets are then combined and used in a supervised classifier to predict the sentiments of customers. We use an audio speech dataset prepared from Amazon product reviews and downloaded from the YouTube portal for the purposes of our experimental evaluations.

• 한국정보과학회논문지:컴퓨팅의 실제 및 레터
• /
• 제14권3호
• /
• pp.336-340
• /
• 2008

최근 감정 분류에 대한 관심이 높아져 연구가 활발히 진행되고 있다. 문서 전체에 관한 감정의 분류도 중요하지만, 문서를 이루고 있는 문장에 관한 분류도 점차 그 필요성이 높아지고 있다. 본 논문에서는 한국어 감정 분류 시스템 구축을 위해서 추출된 한국어 감정 자질을 이용한 한국어 문장 및 문서 감정 분류에 관해 연구한다. 한국어 감정 분류의 시작은 감정을 내포한 대표적인 어휘로부터 시작하며, 이와 같은 감정 자질들은 문장 및 문서의 감정을 분류하는데 결정적인 관여를 한다. 한국어 감정 자질의 추출을 위하여 영어 단어 시소러스 정보를 이용하여 자질들을 확장하고, 영한사전을 통해 확장된 자질들을 번역함으로써 감정 자질들을 추출하였다. 추출된 감정 자질들을 사용하여, 단어 벡터로 표현된 입력문서를 이진 분류기인 지지벡터 기계(SVM: Support Vector Machine)를 이용하여 문장과 문서에 내포된 감정을 판단하고 평가하였다.

• 통합자연과학논문집
• /
• 제14권2호
• /
• pp.35-40
• /
• 2021

In this paper, we proposed a multi-subject-level sentiment analysis model for user reviews using the Latent Dirichlet Allocation (LDA) method targeting user-generated content (UGC). Data were collected from users' online reviews of hotels in major tourist cities in the world, and 30 hotel-related topics were extracted using the entire user reviews through the LDA technique. Six major hotel-related themes (Cleanliness, Location, Rooms, Service, Sleep Quality, and Value) were selected from the extracted themes, and emotions were evaluated for sentences corresponding to six themes in each user review in the proposed sentiment analysis model. Sentiment was analyzed using a dictionary. In addition, the performance of the proposed sentiment analysis model was evaluated by comparing the emotional values for each subject in the user reviews and the detailed scores evaluated by the user directly for each hotel attribute. As a result of analyzing the values of accuracy and recall of the proposed sentiment analysis model, it was analyzed that the efficiency was high.

• Journal of Information Processing Systems
• /
• 제19권6호
• /
• pp.842-857
• /
• 2023

Currently, most sentiment classification models on microblogging platforms analyze sentence parts of speech and emoticons without comprehending users' emotional inclinations and grasping moral nuances. This study proposes a hybrid sentiment analysis model. Given the distinct nature of microblog comments, the model employs a combined stop-word list and word2vec for word vectorization. To mitigate local information loss, the TextCNN model, devoid of pooling layers, is employed for local feature extraction, while BiLSTM is utilized for contextual feature extraction in deep learning. Subsequently, microblog comment sentiments are categorized using a classification layer. Given the binary classification task at the output layer and the numerous hidden layers within BiLSTM, the Tanh activation function is adopted in this model. Experimental findings demonstrate that the enhanced TextCNN-BiLSTM model attains a precision of 94.75%. This represents a 1.21%, 1.25%, and 1.25% enhancement in precision, recall, and F1 values, respectively, in comparison to the individual deep learning models TextCNN. Furthermore, it outperforms BiLSTM by 0.78%, 0.9%, and 0.9% in precision, recall, and F1 values.

• Journal of Information Processing Systems
• /
• 제17권4호
• /
• pp.818-833
• /
• 2021

To address the problems of low precision rate, insufficient feature extraction, and poor contextual ability in existing text sentiment analysis methods, a mixed model account of a CNN-BiLSTM-TE (convolutional neural network, bidirectional long short-term memory, and topic extraction) model was proposed. First, Chinese text data was converted into vectors through the method of transfer learning by Word2Vec. Second, local features were extracted by the CNN model. Then, contextual information was extracted by the BiLSTM neural network and the emotional tendency was obtained using softmax. Finally, topics were extracted by the term frequency-inverse document frequency and K-means. Compared with the CNN, BiLSTM, and gate recurrent unit (GRU) models, the CNN-BiLSTM-TE model's F1-score was higher than other models by 0.0147, 0.006, and 0.0052, respectively. Then compared with CNN-LSTM, LSTM-CNN, and BiLSTM-CNN models, the F1-score was higher by 0.0071, 0.0038, and 0.0049, respectively. Experimental results showed that the CNN-BiLSTM-TE model can effectively improve various indicators in application. Lastly, performed scalability verification through a takeaway dataset, which has great value in practical applications.

• 한국지능시스템학회논문지
• /
• 제23권4호
• /
• pp.317-324
• /
• 2013

트위터 감성 분석은 트윗글의 감성을 긍정과 부정으로 분류하는 작업이다. 이 연구에서는 SentiWordNet(SWN) 감성 사전에 기반한 트윗글 감성 분석을 다룬다. SWN은 전체 영어 단어에 대해 단어의 의미별로 긍정, 부정의 감성 강도를 저장해 둔 감성 사전이다. 기존 SWN 기반 감성 분석 연구들은 문서에 출현하는 각 용어의 감성을 SWN으로부터 결정한 다음 이를 바탕으로 문서 전체의 감성을 결정하였는데, 그 방법들이 매우 다양하다. 예를 들어, 한 용어의 감성 결정 시 해당 용어의 SWN 내 의미별 긍정, 부정 감성 강도 차이들의 평균을 계산하거나 긍정과 부정 각각의 감성 강도 평균 혹은 최대값을 구하기도 하며, 문서 전체의 감성을 결정하는 경우에도 문서 내 용어들의 감성 값들에 대해 평균 혹은 최대값을 취하기도 하였다. 또한 SWN 내 형용사, 동사, 명사, 부사의 품사 집합 전체 혹은 특정 부분집합에 대해 위의 감성 결정 작업을 적용하기도 한다. 이처럼 기존 연구에서는 SWN 기반의 다양한 감성 자질 추출 절차가 시도되고 있으나 이들 자질 추출 기법 전반에 대한 성능 비교 연구는 찾기 힘들다. 이 연구에서는 SWN을 트위터 감성 분석에 활용하는 다양한 방법들을 일반화하는 절차들을 소개하고 각 방법들의 성능 비교 및 분석 결과를 제시한다.

• 한국멀티미디어학회논문지
• /
• 제17권6호
• /
• pp.671-680
• /
• 2014

Recently, with the remarkable increase of social network services, it is necessary to extract interesting information from lots of data about various individual opinions and preferences on SNS(Social Network Service). The sentiment information can be applied to various fields of society such as politics, public opinions, economics, personal services and entertainments. To extract sentiment information, it is necessary to use processing techniques that store a large amount of SNS data, extract meaningful data from them, and search the sentiment information. This paper proposes an efficient method to extract sentiment information from various unstructured big data on social networks using HDFS(Hadoop Distributed File System) platform and MapReduce functions. In experiments, the proposed method collects and stacks data steadily as the number of data is increased. When the proposed functions are applied to sentiment analysis, the system keeps load balancing and the analysis results are very close to the results of manual work.