• Proceedings of the Korea Information Processing Society Conference
• /
• 2014.11a
• /
• pp.905-908
• /
• 2014

악성댓글은 인터넷 상에서 상대방이 올린 글에 대한 비방, 험담 등을 하는 악의적인 댓글을 의미한다. 사용자에게 스마트 모바일 기기, 소셜 네트워크 서비스 등의 편리한 서비스를 제공함에 따라 악성댓글에 대한 피해도 꾸준히 증가하고 있다. 본 논문에서 제안하는 방법은 댓글로부터 간단한 형태소 분석과 패턴 추출 과정을 거쳐 단어장을 형성한다. 단어장을 바탕으로 댓글에 포함된 단어가 악성댓글과 비악성댓글에서 나타날 확률을 구하고 이를 기반으로 주어진 댓글이 악성댓글인지 아닌지를 판별한다. 실험결과를 통하여 본 논문에서 제안하는 악성댓글을 판별하는 방법을 평가한다.

• Asia-pacific Journal of Multimedia Services Convergent with Art, Humanities, and Sociology
• /
• v.8 no.12
• /
• pp.817-826
• /
• 2018

As malicious comments are emerging as social problems, the solution is needed. Many studies have been conducted in various perspectives to understand and prevent malicious comments. In the previous researches, the neutralization of malicious comments has attracted attention as an important factor explaining the malicious comments, but the difference of the degree of neutralization according to the gender has not been rarely considered. In addition, although there are many environmental characteristics that are different from reality in online, research with malicious comments is insufficient. Based on these facts, this study examined moderating effects of gender on relationship between malicious comments and neutralization, and demonstrated the effects of online environmental factors (anonymity, lack of social presence) on malicious comments. As a result of the study, we discovered that the influence of online environmental factors was not found, but neutralization of malicious comments had strong direct influence on malicious comments and moderating effect of gender difference. Based on the results of this study, we discuss academic and practical implications and suggest limitations of research and future research directions.

• Smart Media Journal
• /
• v.11 no.2
• /
• pp.18-24
• /
• 2022

Problems caused by malicious comments occur on many social media. In particular, YouTube, which has a strong character as a medium, is getting more and more harmful from malicious comments due to its easy accessibility using mobile devices. In this paper, we designed and implemented a YouTube malicious comment detection system to identify malicious comments in YouTube contents through LSTM-based natural language processing and to visually display the percentage of malicious comments, such commentors' nicknames and their frequency, and we evaluated the performance of the system. By using a dataset of about 50,000 comments, malicious comments could be detected with an accuracy of about 92%. Therefore, it is expected that this system can solve the social problems caused by malicious comments that many YouTubers faced by automatically generating malicious comments statistics.

• Journal of Digital Convergence
• /
• v.16 no.12
• /
• pp.197-206
• /
• 2018

The harmful effects of online malicious comments are continuously increasing. Many previous studies have confirmed that neutralization of malicious comments is a key predictor. Neutralization is theoretically composed of seven multidimensional concepts, and the significance of neutralization factors varies depending on the type of deviant behavior. This study focuses on the fact that the malicious comment researches have considered the neutralization techniques in a single dimension as opposed to demonstrating the multidimensional neutralization techniques in the deviant behavior research. On the other hand, the role of arbitrator in deviant behavior can contribute to restraining deviant behavior, but the research of intervention intention is relatively lacking in malicious comments research. This study, composed of two complementary studies, tried to find out the related factors of malicious comments and intervention intention. As a result of study, This study revealed that malicious commentator uses the neutralization techniques of condemn the condemners and denial of responsibility. In addition, we found that affective empathy has a significant effect on the intervention intention in malicious comments.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.11a
• /
• pp.775-778
• /
• 2021

악성 댓글은 언어폭력이며 사이버 범죄의 일종으로 인터넷상에서 상대방이 올린 글에 비방이나 험담을 하는 악의적인 댓글을 말한다. 악성 댓글을 단순히 차단하는 다른 프로그램들과는 달리 해당 영상의 악성 댓글의 비율을 알려주고 악플러들의 닉네임과 그 빈도를 나타내주는 것으로 차별화를 두었다. 따라서 많은 유튜버들이 겪는 악성 댓글 문제들을 탐지하여 유튜브에 달리는 악성 댓글들을 탐지하고 시각화하여 제공한다.

• Annual Conference on Human and Language Technology
• /
• 2015.10a
• /
• pp.178-180
• /
• 2015

본 논문에서는 기계 학습(Machine Learning)을 이용하여 댓글의 악성 여부를 분류하는 시스템에 대해 설명한다. 댓글은 문장의 길이가 짧고 맞춤법이 잘 되어있지 않는 특성을 가지고 있다. 따라서 댓글 분석을 위해 형태소 분석 결과와 문자단위 Bi-gram, Tri-gram을 자질로 이용한다. 전처리 된 댓글에서 각 자질 추출 방법에 따라 자질을 추출한다. 추출된 자질을 이용하여 기계학습 알고리즘의 모델을 학습하고 댓글의 악성 여부 분류에 활용한다. 본 논문에서는 댓글의 악성 여부 판별을 위한 자질 추출방법을 제안하고 실험을 통해 이에 대한 효용성을 검증하였다.

• 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.

• Journal of the Korea Convergence Society
• /
• v.11 no.6
• /
• pp.23-28
• /
• 2020

A comment system is essential for communication on the Internet. However, there are also malicious comments such as inappropriate expression of others by exploiting anonymity online. In order to protect users from malicious comments, classification of malicious / normal comments is necessary, and this can be implemented as text classification. Text classification is one of the important topics in natural language processing, and studies using pre-trained models such as BERT and graph structures such as GCN and GAT have been actively conducted. In this study, we implemented a comment classification system using BERT, GCN, and GAT for actual published comments and compared the performance. In this study, the system using the graph-based model showed higher performance than the BERT.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.16 no.9
• /
• pp.2043-2049
• /
• 2012

Because of the growth of internet and development of smart phone technology, peoples are able to connect the internet easily and convenient at anytime, and anywhere. The other hand, that take place many serious social side-effect and one of a typical cases is the problem of malicious replay in weaken of netizen's internet ethics. The problem of malicious replay is increasing more and more the seriousness of the problem and the victims are suffering from much the pains of the trauma by a malicious replay. Therefore, in this paper, we investigate and analyze actual on aware of internet ethics and malicious replay for university students and then we propose the improvement of the problem.

• Annual Conference on Human and Language Technology
• /
• 2023.10a
• /
• pp.577-580
• /
• 2023

악성댓글은 인터넷상에서 정서적, 심리적 피해를 주는 문제로 인식되어 왔다. 본 연구는 한국어 악성댓글 탐지 분석을 위해 KcBERT 및 다양한 모델을 활용하여 성능을 비교하였다. 또한, 공개된 한국어 악성댓글 데이터가 부족한 것을 해소하기 위해 기계 번역을 이용하고, 다국어 언어 모델(Multilingual Model) mBERT를 활용하였다. 다양한 실험을 통해 KcBERT를 미세 조정한 모델의 정확도 및 F1-score가 타 모델에 비해 의미 있는 결과임을 확인할 수 있었다.