DOI QR코드

DOI QR Code

Feature Extraction to Detect Hoax Articles

낚시성 인터넷 신문기사 검출을 위한 특징 추출

  • 허성완 (아주대학교 소프트웨어특성화학과) ;
  • 손경아 (아주대학교 소프트웨어학과)
  • Received : 2016.03.15
  • Accepted : 2016.08.09
  • Published : 2016.11.15

Abstract

Readership of online newspapers has grown with the proliferation of smart devices. However, fierce competition between Internet newspaper companies has resulted in a large increase in the number of hoax articles. Hoax articles are those where the title does not convey the content of the main story, and this gives readers the wrong information about the contents. We note that the hoax articles have certain characteristics, such as unnecessary celebrity quotations, mismatch in the title and content, or incomplete sentences. Based on these, we extract and validate features to identify hoax articles. We build a large-scale training dataset by analyzing text keywords in replies to articles and thus extracted five effective features. We evaluate the performance of the support vector machine classifier on the extracted features, and a 92% accuracy is observed in our validation set. In addition, we also present a selective bigram model to measure the consistency between the title and content, which can be effectively used to analyze short texts in general.

스마트 기기의 발달로 많은 사람들이 인터넷 신문기사를 이용하고 있다. 하지만 인터넷 언론사간의 치열한 경쟁으로 조회수를 올리기 위한 낚시성 기사가 범람하고 있다. 낚시성 신문기사는 제목을 통해 올바른 기사의 줄거리가 제공되지 않았을 뿐만 아니라, 독자로 하여금 잘못된 내용을 떠올리게 한다. 낚시성 신문기사는 핵심에서 벗어난 유명인사 인용, 애매한 문장의 마무리, 제목과 내용의 불일치 등의 특징을 갖는다. 본 논문에서는 이러한 낚시성 기사를 분류하기 위한 특징을 추출하고 성능을 검증해 본다. 기사에 달린 댓글의 키워드를 활용하여 대용량 학습데이터를 생성하고 이를 기반으로 다섯 가지 분류 특징을 추출하였다. 추출된 특징들은 서포트 벡터 머신 분류기를 이용한 실험에서 92%의 정확도를 보여 낚시성 인터넷 신문기사를 분류하는데 적합하다고 판단된다. 뿐만 아니라 제목과 본문의 일관성을 측정하기 위한 전처리 방법으로 고안한 선택적 바이그램 모델은 낚시성 인터넷 신문기사 분류 외에도 일반적인 단문 분석을 위한 전처리 방법으로 유용할 것으로 기대된다.

Keywords

Acknowledgement

Supported by : 한국연구재단

References

  1. Korea internet security agency(KISA), "2015 Survey on the Internet usage executive (in Korean)," 2015.
  2. Statistics Korea, "2015 Social Survey results (in Korean)," 2015.
  3. Kin sunjin, "The utilization reality of hooking news titles in portal news service - Focused on major daily newspapers in naver newscast(in Korean)," Korea Digital Design Council, Study on digital design, Vol. 10, No. 4, pp. 283-293, 2010.
  4. Jeong kyihong, "Is internet hoax articles good as it is?" (in Korean), Kwanhun Club, Journal of Kwanhun, 115, pp. 129-141, 2010.
  5. Lim Kwon-Mook, "Morphological Ambiguity Resolution by using Semantic Information(in Korean)," Korean Institute of Information Scientists and Engineers, Conference Proceedings, Vol. 21, pp. 649-652, 1994.
  6. eunjeonhanip project, [Online]. Available: http://eunjeon.blogspot.kr/
  7. Lee joon-ho, "Implications and Theories of Headline Journalism : Focusing on Headlines of Newspapers and Portal Sites(in Korean)," Society for journalism and communication studies, Journalism Research, pp. 249-280, 2015.
  8. Jang min-jeong, "Analysis of newspaper headlines among Korea, USA and Japan (in Korean)," International Association of Language & Literature Conference Proceedings, pp. 25-37, 2012.
  9. Janggeun Oh, "Semiotic analysis of the daily press Edit(in Korean)," Korean Association for Visual Culture, Visual culture, pp. 108-121, 2007.
  10. Rhee, June Woong, "Direct Quotations in newspaper headlines in the coverage of the local election on May 31, 2006," Korea Journal of Jounalism & Communication studies, pp. 64-90, 2007.
  11. Yu, Hong-Sik, "Effects of Quantitative Distortion in the Equalized Exemplification on Readers' Issue Perception(in Korean)," Korea Journal of Jounalism & Communication studies, pp. 346-373, 2008.
  12. Corinna cortes, "Support-vector networks," Machine Learning, Vol. 20, Issue 13, pp. 273-297, 1995.