An Empirical Comparison of Machine Learning Models for Classifying Emotions in Korean Twitter

Abstract

As online texts have been rapidly growing, their automatic classification gains more interest with machine learning methods. Nevertheless, comparatively few research could be found, aiming for Korean texts. Evaluating them with statistical methods are also rare. This study took a sample of tweets and used machine learning methods to classify emotions with features of morphemes and n-grams. As a result, about 76% of emotions contained in tweets was correctly classified. Of the two methods compared in this study, Support Vector Machines were found more accurate than Naïve Bayes. The linear model of SVM was not inferior to the non-linear one. Morphological features did not contribute to accuracy more than did the n-grams.

Text Mining and Sentiment Analysis for Predicting Box Office Success

“…By the way, sentiment analysis is often used in opinion mining which focuses on extracting authors' opinions and sentiments from user-generated content such as customer reviews, forum messages, and blogs [1], [12], [13]. Methods for extracting sentiment and opinion from text can be divided into the following two approaches: the lexicon-based approach and the machine learning approach [10]. The machine learning approach generates classifiers by training data set with labels and the lexicon-based approach applies a linguistic resource called a sentiment dictionary or sentiment lexicon.…”

“…Despite of the research, however, general dictionaries have been preferred than manual sentiment dictionaries because of their high coverage and reliability [18]. Besides, Korean word-ofmouth data has seldom been examined in sentiment analysis because Korean is an agglutinative language, which makes it difficult to analyze morphemes within corpora [10].…”

“…To assess the classification performance, we measured recall, precision, accuracy α , and F1-scores [10], [19], [20]. Total accuracy α is defined as the percentage of sentiments correctly predicted of the total instances.…”

“…However the studies have inspired us to develop a sentiment lexicon for better performance in emotion mining, there are rarely references to tools, techniques, and methods to build a domain-specific sentiment dictionary. Besides, Korean word-ofmouth data has seldom been examined in sentiment analysis because Korean alphabet "Hangeul" is an agglutinative language, which makes it difficult to analyze morphemes within corpora [10]. Therefore, we tried to investigate the usability of emotion mining for business decision making, and targeted the Korean film market, box office success of movies is seriously affected by electronic word-of-mouth.…”

Investigations on Techniques and Applications of Text Analytics

“…Learning)을 통해 감정의 극성을 분석한 연구 [105] , 특 정 감정을 통해 다우 존스 평균 주가의 예측을 시도한 연구 [106] , 동일한 데이터에 SVM 기법과 나이브 베이 즈 기법을 적용한 후 그 결과를 비교한 연구 [107] 등이 있다. 트위터 외에도 감성 분석은 블로그 데이터 [108][109] , 뉴스 데이터 [110] , 인스타그램 [111] , 페이스북 [112] 그리고 영화 리뷰 [91,[113][114] [115] , 뉴스, 블로그, 트위터 데이터를 대 상으로 의미적 요소의 결합을 시도한 연구 [116] , 그리고 뉴스, 블로그, 트위터 데이터에 감성 분석을 통해 선 호 농식품 리스트를 추천한 연구 [117] 를 들 수 있다.…”