A Study on Social Issues Automatic Detection Methods in Dynamic Social Communities

Abstract

As the main agent is changed from supplier to user, social media is spotlighted. The early media users provided information to share information using computer, but propagating of smart phone, the form has been changed that users provided information through portable devices such as smart phone and tablet PC. Providing information through portable devices simplified the message form of information delivery, such form occurred popularity of social media service providing simple information delivery. There’re spotlighted social media such as Twitter, Facebook, Instagram and so on. To these media, a research was being carried out to utilize posted information as posting news or individual interests becoming the headlines. Progressing researches are that interest detection of social media users, issue tracking, trend detection and user analyzation. And the most researches are being carried out relating to topic in social media, so topic detection stood out as important research. There are topic modeling algorithms such as LDA, ESA, LSA, HDP and NMF in topic detection description methods. And researches using these algorithms are focused on topic detection, so there’re ambiguous drawbacks that analyzation which the topic is used in which field to be issue. To solve this drawbacks, in this paper, twits were collected to detect social issues in twitter, and the twits were recasted to apply to topic modeling. After that, using topic modeling, the topic group is extracted and it’s recommended that classifying method that social issues of the group using news retrieving about vocabularies constituting the extracted topic. Through recommended method, the social issue is detected and it’s classified that the issue is included to any field in twitter. Through classified social issue like this, users’ tendencies and interests are grasped and it’s utilized to recommending system and prediction system development.