• Journal of Advanced Marine Engineering and Technology
• /
• v.34 no.6
• /
• pp.871-879
• /
• 2010

In this paper, we present a system that classifies comments on a news article into a user opinion called a polarity (positive or negative). The system is a kind of document classification system for comments and is based on machine learning techniques like support vector machine. Unlike normal documents, comments have their body that can influence classifying their opinions as polarities. In this paper, we propose a feature weighting scheme using such characteristics of comments and several resources for opinion classification. Through our experiments, the weighting scheme have turned out to be useful for opinion classification in comments on Korean news articles. Also Korean character n-grams (bigram or trigram) have been revealed to be helpful for opinion classification in comments including lots of Internet words or typos. In the future, we will apply this scheme to opinion analysis of comments of product reviews as well as news articles.

• Journal of Industrial Convergence
• /
• v.20 no.12
• /
• pp.135-143
• /
• 2022

This paper aims to grasp what propensity users have in their ideology from the comments in the Web Portal News and the Press Website News. Through these analytical results, the political propensities of not only the Web Portal News and the Press Website News but also the voters who use these news media could be grasped. The collection of data necessary for this study has been made from the comments of 174 news stories for about 90 days before the election day. For the analysis, T-test has been used in order to compare Naver News with Daum News, the Minjoo Party of Korea with the People Power Party, and the Press Web Site News with Naver News. As a result of the analysis, the comments of Naver News took the higher percentage in the positive writings about the candidates of the conservative party. but, in contrast, those of Daum News in that percentage were higher about the ones of the progressive party. Accordingly, it can be found that Naver News is mainly used by users with the politically conservative propensity, while Daum News is mostly used by those with progressive one.

• Korean journal of communication and information
• /
• v.79
• /
• pp.165-195
• /
• 2016

This paper aims to examine the third-person effect(TPE) of hate comments on online news, and analyze how the issue-relevant audience factors as well as the characteristics of the online message have influence on the TPE. More specifically, based on the distinction between hateful and logical comments regarding the issue of illegal immigration, the authors have conducted an online experiment that compares how the message-related features, i.e., ways of expressing the ideas, lead to the difference in TPE. Analysis was also conducted with regards to how political orientation and discriminatory predisposition to immigrants among the audiences, have different impacts on the TPE. The 479 participants in the experiments were randomly assigned to experimental group(exposed to hate comments) or control group(exposed to logical comments). The results reveal that the TPE of hate comments is higher than that of logical message. The same message proved to be more effective for news users with liberal orientation and discriminatory predisposition. The significance of this paper lies in that it has examined the effect of online hate comments in a rigorous experimental setting. Also the research further elaborated on the audience-related variables, for which the previous studies tended to focus those on the general psychological level rather than relate them more specifically to the issues.

• Korean journal of communication and information
• /
• v.42
• /
• pp.44-78
• /
• 2008

The present study explored the agenda setting effects of replies called "Daet-Gul", and perception of the news replies. This study has established three research questions: 1) the recognition of the online communication 2) the degree of the reading and writing on online spare 3) the amount of the effects on the online communication. This study is performed using survey method. The survey results indicated in that the participants are very passive readers and writers on the online spare. In addition, the survey repliers evaluated that replies' mechanical device and antigravitational speed have high score, whereas they marked low store in the content and credibility of 'the replies. Therefore, they did not estimate the effects of the replies highly. All the results indicate that 'the replies' is not the fundamental factors of the deliberative democracy. It's because online communication with 'the replies' are thought to be fated the abuse and slander. Therefore, it's essential to improve the online communication with 'the replies', through the introduction of the 'trackback', which is a sort of the 'remote replies'

• The Journal of the Korea Contents Association
• /
• v.13 no.12
• /
• pp.201-211
• /
• 2013

This study examined the effect of valence for best reply on the conformity of general reply in online environment. A total of 194 participants participated in this study, each participant assigned randomly in three experimental groups(positive, negative, and control). Participants were asked to read online news article, best reply and general 6 replies, and then, to write their own opinions in the reply section. In addition, the level of self-expression and issue commitment were measured. The contents of reply participants written was categorized three valence(positive, negative, and neutral) by the four experimenters' judgment. The mean of inter-rater reliability was 84.9%. The results indicated that the level of self-expression and issue commitment were comparable across experimental conditions. However, the result of cross-table analysis showed that there is a significant difference in the valence of general reply across experimental conditions. Specifically, there were significant difference in the valence of general reply between positive and negative experimental group and positive and control group, but there is no significant difference between negative and control group.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2009.04a
• /
• pp.226-229
• /
• 2009

최근 들어 인터넷 게시판이나 개인 블로그 등은 온라인상에서 사람들의 정보 공유나 의견 교환의 중요한 매체가 되고 있다. 많은 수의 블로그들은 현재 사회적으로 이슈가 되는 여러 문제들을 반영하고 있다. 또한 최근 댓글을 통해 적극적으로 자신의 의사 표현하거나 다른 사람들의 의견을 살피는 인터넷 사용자의 증가로 인터넷 뉴스나 블로그 기사에 많은 수의 댓글이 달리고 있다. 그러나 대부분의 블로그나 인터넷 포털 사이트의 경우 기사나 댓글들을 순차적인 목록 형태로 제공하므로 자신이 원하는 내용의 댓글을 검색하거나 전체 댓글에 대한 전반적인 파악은 힘든 일이다. 따라서 본 논문에서는 기사에 달린 많은 수의 댓글들을 분류하고, 이를 시각화 하는 시스템인 TRIB(Telescope for Responding comments for Internet Blog)을 제안한다. TRIB은 미리 정의된 사용자 정의 사전을 이용하여 댓글을 내용에 따라 분류하여 시각화 하므로 사용자들은 자신의 관심과 흥미에 따라 개인화 된 뷰를 볼 수 있다. 1,000개 이상의 댓글을 가진 뉴스 기사들을 대상으로 한 실험을 통해 TRIB 시스템의 댓글 분류와 시각화 성능을 보인다.

• The Journal of the Korea Contents Association
• /
• v.19 no.1
• /
• pp.548-558
• /
• 2019

Along generalization of internet news comments, malicious comments have been spread and made many social problems. Because writings reflect human mental state or trait, analyzing malicious comments, human mental states could be inferred when they write internet news comments. In this study, we analyzed malicious comments of English and Korean speaker using LIWC and KLIWC. As a result, in both English and Korean, malicious comments are commonly more used in sentence, word phrase, morpheme, word phrase per sentence, morpheme per sentence, positive emotion words, and cognitive process words than normal comments, and less used in the third person singular, adjective, anger words, and emotional process words than normal comments. This means people are state that they can not control their feeling such as anger and can not think well when they write news comments. Therefore, when internet comments were written, service provider should consider the way that commenters monitor own writings by themselves and that they prevent the other users from getting close to comments included many negative-emotion words. In other sides, it is discovered that English and Korean malicious comments was discriminated by authenticity. In order to be more objective, gathering data from various point of time is needed.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.11a
• /
• pp.864-867
• /
• 2021

뉴스, SNS 등의 인터넷 댓글은 익명으로 의견을 자유롭게 개진할 수 있는 반면 댓글의 익명성을 악용하여 비방이나 험담을 하는 악성 댓글이 여러 분야에서 사회적 문제가 되고 있다. 해당 문제를 해결하기 위해 AI를 활용한 댓글 분류 알고리즘을 개발하려는 많은 노력들이 이루어지고 있지만, 댓글 분류 모델에 사용되는 AI는 오버피팅의 문제로 인해 댓글 분류에 대한 정확도가 떨어지는 문제점을 가지고 있다. 이에 본 연구에서는 크라우드소싱을 활용하여 오버피팅으로 인한 악성 댓글 분류 및 판단 정확도 저하 문제를 개선한 크라우드소싱 기반 딥러닝 분류 알고리즘(Deep Learning Classification Algorithm Based on Crowdsourcing: DCAC)과 해당 알고리즘을 사용한 시스템을 제안한다. 또한, 실험을 통해 오버피팅으로 낮아진 판단 정확도를 증가시키는 데 제안된 방법이 도움이 되는 것을 확인하였다.

• Korean journal of communication and information
• /
• v.46
• /
• pp.177-211
• /
• 2009

The purpose of this study is to investigate how portal users recognize the role and function of the portal as a press. Furthermore, this study aims to analyze the pattern of portal news use among portal users and their recognition about replies to news. For this, we conducted online survey of 1,036 people aged from 15 to 45, who used portal site at least once a week. As a result, most users(70.8%) agreed that portal news played a role as a press with other media. In addition, six out of ten agreed that portal news set the agenda of society. These results suggest that portal users think highly of the social impact and responsibility of portal in society. However, about 40 percent of respondents said that they did not trust in information shown in replies to news and many replies to news were written by albeit. Six out of ten thought that replies to news were written by intention. Therefore, to manage the quality of replies to news is necessary to secure the credibility of portal news. As respondents use portal news often, they tend to think that portals are convenient to use and their social impact is very large, at the same time considering their sensation and commercialization. On the other hand, as portal users use the internet often, they tend to think that portals are convenient to use and their social impact is immense. However, there was no significant relationship between income and portal users' recognition of portals as the press.

• Annual Conference on Human and Language Technology
• /
• 2020.10a
• /
• pp.437-440
• /
• 2020

최근 자연어 처리에서는 사전 학습과 전이 학습을 통하여 다양한 과제에 높은 성능 향상을 성취하고 있다. 사전 학습의 대표적 모델로 구글의 BERT가 있으며, 구글에서 제공한 다국어 모델을 포함해 한국의 여러 연구기관과 기업에서 한국어 데이터셋으로 학습한 BERT 모델을 제공하고 있다. 하지만 이런 BERT 모델들은 사전 학습에 사용한 말뭉치의 특성에 따라 이후 전이 학습에서의 성능 차이가 발생한다. 본 연구에서는 소셜미디어에서 나타나는 구어체와 신조어, 특수문자, 이모지 등 일반 사용자들의 문장에 보다 유연하게 대응할 수 있는 한국어 뉴스 댓글 데이터를 통해 학습한 KcBERT를 소개한다. 본 모델은 최소한의 데이터 정제 이후 BERT WordPiece 토크나이저를 학습하고, BERT Base 모델과 BERT Large 모델을 모두 학습하였다. 또한, 학습된 모델을 HuggingFace Model Hub에 공개하였다. KcBERT를 기반으로 전이 학습을 통해 한국어 데이터셋에 적용한 성능을 비교한 결과, 한국어 영화 리뷰 코퍼스(NSMC)에서 최고 성능의 스코어를 얻을 수 있었으며, 여타 데이터셋에서는 기존 한국어 BERT 모델과 비슷한 수준의 성능을 보였다.