• Annual Conference of KIPS
• /
• 2013.05a
• /
• pp.1038-1039
• /
• 2013

최근 스마트폰의 사용이 보편화되면서 많은 양의 온라인 뉴스가 다양한 경로를 통하여 서비스되고 있다. 한편, 실시간으로 제공되는 뉴스의 양이 방대해지면서, 언론사에서 톱 뉴스로 제공하는 토픽과 달리, 실제 사용자들에게 화제가 되고 있는 토픽을 선별하는 데 어려움이 있다. 많은 사용자들이 실생활에서 작성하고 공유하는 트위터는 실제 사람들 사이에 화제가 되고 있는 토픽을 담고 있는 경우가 많다. 이러한 트렌드를 뉴스와 연계시키면 화제가 되는 트렌드 뉴스를 사용자에게 제공할 수 있다. 본 논문에서는 클라이언트-서버 모델을 기반으로 실시간으로 사용자 트위터를 분석하여 추출된 트렌드를 기반으로 관련 뉴스를 검색하여 제공하는 시스템을 제안한다. 클라이언트를 통해 수집한 트위터 단문에서 서버는 화제가 되고 있는 트렌드를 추출하고, 이를 기반으로 Google 등을 통해 관련 뉴스를 검색하여 클라이언트에게 전달한다. 이 모든 과정을 실시간으로 제공하기 위한 알고리즘을 제안하고 프로토타입 시스템을 통하여 그 성능을 평가한다.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.43 no.6
• /
• pp.811-819
• /
• 2023

Understanding the local conditions is a crucial factor in enhancing the success potential of overseas construction projects. This can be achieved through the analysis of news articles of the target market using topic modeling techniques. In this study, the authors aimed to analyze news articles using two topic modeling methods, namely Latent Dirichlet Allocation (LDA) and BERTopic, in order to determine the optimal approach for market condition analysis. To evaluate the alignment between the generated topics and the actual themes of the news documents, the research collected 6,273 BBC news articles, created ground truth data for individual news article topics, and finally compared this ground truth with the results of the topic modeling. The F1 score for LDA was 0.011, while BERTopic achieved a score of 0.244. These results indicate that BERTopic more accurately reflected the actual topics of news articles, making it more effective for understanding the overseas construction market.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2019.11a
• /
• pp.11-13
• /
• 2019

대중들에게 정보를 빠르고 정확하게 제공하는 대표 매체인 뉴스 기사는 일 평균 1만 5천 건 이상이 보도되고 있다. 특정 주제 또는 분야에 대한 전반적인 동향을 파악하고자 대량의 텍스트 데이터를 수집하여 텍스트 마이닝(Text mining)과 머신러닝 등을 적용하는 연구들이 활발하게 수행되고 있다. 본 연구에서는 서울시의 이슈 및 문제를 파악하고자 약 5년간 뉴스 기사를 수집하여 키워드 분석 및 토픽 모델링을 적용하였다. 분석 결과 5년간의 뉴스 기사에서 빈번하게 출현하는 키워드들을 도출하였고 연도별로 도출된 키워드들을 비교분석하였다. 또한 토픽 모델링 적용 결과 뉴스 기사를 구성하는 20개의 주제를 도출하였으며 이를 기반으로 서울시의 주요 이슈들을 파악할 수 있다. 본 연구는 연도별, 분야별 세부 내용 및 시계열 분석, 다른 도시들의 이슈 및 문제를 도출하는데 활용될 것으로 기대된다.

• Journal of Digital Convergence
• /
• v.18 no.3
• /
• pp.1-12
• /
• 2020

The purpose of this study is to explore the trend of blockchain technology through analysis of patents and news articles using text mining, and to suggest the blockchain policy agenda by grasping social interests. For this purpose, 327 blockchain-related patent abstracts in Korea and 5,941 full-text online news articles were collected and preprocessed. 12 patent topics and 19 news topics were extracted with latent dirichlet allocation topic modeling. Analysis of patents showed that topics related to authentication and transaction accounted were largely predominant. Analysis of news articles showed that social interests are mainly concerned with cryptocurrency. Policy agendas were then derived for blockchain development. This study demonstrates the efficient and objective use of an automated technique for the analysis of large text documents. Additionally, specific policy agendas are proposed in this study which can inform future policy-making processes.

• Korean journal of communication and information
• /
• v.30
• /
• pp.175-201
• /
• 2005

This study investigates the diversity of news topic and source of the New York Times, Chicago Tribune, and the Los Angeles Times in the second half of the twentieth century. In probing the conventional traits of the contemporary press, the researcher traced the changing patterns and trends of news values in terms of news-gathering routine in order to evaluate the journalistic role conception in terms of social responsibility theory. Findings indicated that the American press as a neutral transmitter has been consistently violated by source and topic bias without any significant changes during the last five decades. The data, however, revealed the evident shift of the contemporary press from the heavy reliance of official source to the business/economic source. In addition, news topics such as business, health, and education have replaced the conventional popular topics such as crime and accidents. By contrast, it was revealed that the unconventional topics such as poverty, labor and minority still fail to receive the large attention from the target papers.

• The Journal of the Korea Contents Association
• /
• v.20 no.5
• /
• pp.457-466
• /
• 2020

The purpose of this study is to find out what the main agenda of social formation is and how it changes through the media by utilizing the news big data of COVID-19 which is spreading recently, and to suggest the direction of future reporting. In order to achieve the purpose of the research, 47,816 cases of news big data reported from December 31, 2019 to March 11, 2020 were divided into four periods based on the fourth stage of the crisis warning for infectious diseases, and a total of 20 topics were derived. Based on the results of the Topic Modeling analysis, this study proposed the following. First, it is necessary to refrain from provocative expressions such as "anxiety" and "fear" and use neutral and objective reporting terms. Second, more in-depth and contextual news production is required, breaking away from simple event news production. Third, it is necessary to prepare detailed crisis communication manuals for each situation related to infectious diseases. Fourth, we need reports that focus on citizens-led efforts to overcome the crisis. This research has the academic significance that it is the first paper to analyze news big data on COVID-19 using the Topic Modeling Analysis method, and the policy significance that can be used as the basis for developing national crisis communication policy.

• Journal of Digital Convergence
• /
• v.19 no.11
• /
• pp.575-583
• /
• 2021

This study aims to analyze Taekwondo trends according to news articles by year by applying topic modeling. In order to examine the Taekwondo trend through media reports, articles including news articles and Taekwondo specialized media articles were collected through Big Kinds of the Korea Press Foundation. The search period was divided into three sections: before 2000, 2001~2010, and 2011~2020. A total of 12,124 items were selected as research data. For topic analysis, pre-processing was performed, and topic analysis was performed using the LDA algorithm. In this case, python 3 was applied for all analysis. First, as a result of analyzing the topics of media articles by year, 'World' was the most common keyword before 2000. 'South and North Korea' was next common and 'Olympic' was the third commonest topic. From 2001 to 2010, 'World' was the most common topic, followed by 'Association' and 'World Taekwondo'. From 2011 to 2020, 'World', 'Demonstration', and 'Kukkiwon' was the most common topic in that order. Second, as a result of analyzing news articles before 2000 by topic modeling, topics were divided into two categories. Specifically, Topic 1 was selected as 'South-North Korea sports exchange' and Topic 2 was selected as 'Adoption of Olympic demonstration events'. Third, as a result of analyzing news articles from 2001 to 2010 by topic modeling, three topics were selected. Topic 1 was selected as 'Taekwondo Demonstration Performance and Corruption', Topic 2 was selected as 'Muju Taekwondo Park Creation', and Topic 3 was selected as 'World Taekwondo Festival'. Fourth, as a result of analyzing news articles from 2011 to 2020 by topic modeling, three topics were selected. Topic 1 was selected as 'Successful Hosting of the 2018 Pyeongchang Winter Olympics', Topic 2 was selected as 'North-South Korea Taekwondo Joint Demonstration Performance', and Topic 3 was selected as '2017 Muju World Taekwondo Championships'.

• The Journal of the Korea Contents Association
• /
• v.21 no.6
• /
• pp.218-227
• /
• 2021

The purpose was to examine what kind of discourse about foreigners in the media in a situation where hatred toward foreigners prevailed in a pandemic situation. News data related to Chinese international students(CIS) was collected for 2020, The 11 optimal topics were selected derived through LDA analysis. They were analyzed in an exploratory level, focusing on the relationship with major events per year. The news about CIS in 2020 was intensively linked to reports on the COVID19 situation. There was a tendency to report in response to the presupposes CIS as potential confirmed patients.

• The Journal of the Korea Contents Association
• /
• v.22 no.7
• /
• pp.30-45
• /
• 2022

This study explored the issues of industrial mathematics in online news articles and online forums in Korea and the US by using text mining and compared the results. Text data about industrial mathematics were collected from news articles of Naver, a major portal site, and postings and replies on Clien as resources of Korea, and from news articles by the New York Times and CNN as well as postings and replies on Reddit as resources of the US. Structural topic modeling analyses were performed, the major results of which were as follows. First, news articles in Korea mainly dealt with the necessity of industrial mathematics and government support. On the contrary, the news articles in the US focused more on various fields where industrial mathematics fields were utilized. Second, in Korea, the same number of issues with different topics were discussed in news articles and online forums, whereas in the US more issues were covered in news articles than in online forums. It was suggested academic implications for researchers and practical implications for the government for settling industrial mathematics in Korea.

• Annual Conference on Human and Language Technology
• /
• 2021.10a
• /
• pp.287-292
• /
• 2021

좋은 자연어 이해 시스템은 인간과 같이 텍스트에서 단순히 단어나 문장의 형태를 인식하는 것 뿐만 아니라 실제로 그 글이 의미하는 바를 정확하게 추론할 수 있어야 한다. 이 논문에서 우리는 뉴스 헤드라인으로 뉴스의 토픽을 분류하는 open benchmark인 KLUE(Korean Language Understanding Evaluation)에 대하여 기존에 비교 실험이 진행되지 않은 시중에 공개된 다양한 한국어 라지스케일 모델들의 성능을 비교하고 결과에 대한 원인을 실증적으로 분석하려고 한다. KoBERT, KoBART, KoELECTRA, 그리고 KcELECTRA 총 네가지 베이스라인 모델들을 주어진 뉴스 헤드라인을 일곱가지 클래스로 분류하는 KLUE-TC benchmark에 대해 실험한 결과 KoBERT가 86.7 accuracy로 가장 좋은 성능을 보여주었다.