SNS 텍스트 데이터 기반 대전시 환경 관련 토픽 모델링 및 감성 분석

Abstract

The purpose of this study is to analyze how the types of posting or topics that appear on specific topics on portal sites are composed and what differences occur in emotional vocabulary used in relation to them. Especially, in relation to the environment, the topics of content posted on SNS are identified, and the relationship between the topics of these posts and sentiment words sre identified. For this study, topic modeling (LDA) and sentiment analysis on SNS text data, focusing on the keywords of ‘Daejeon’, a large city, and environment'. First, social media data, including the topic and content of postings was collected from blogs, cafes, web documents, and Daejeon local newspapers on portal sites and then preprocessed. Using Python, modeling was performed on the topic of the post, and it was verified whether there was a difference in the emotional vocabulary used for each topic. Using Python, modeling was performed on the topic of the post, and it was tested whether there was a difference in the sentimental word used for each topic. In 2018, it became a big social issue when it became known that radon, a radioactive material, was detected above the standard in a bed sold by a bed company. In addition, as environmental risk factors such as fine dust, asbestos, and air pollution are rapidly increasing, people's interest in the environment is increasing, which is the background of this study. As a result of this study, it was confirmed that there is a significant difference in the positive and negative sentiment words used depending on the topic of the content. It was also interesting to gain a new insight that people are interested in 'healthy food' along with the environment.