Keyword Analysis and Topic Extraction of Hospital Violence News

摘要

Currently disputes between doctors and patients have increased in China, as has hospital violence. How to improve the relationship between them is an important problem. This paper focuses on the online news about hospital violence from 2013 to 2018. If we can find out the characteristics of hospital violence news, we will have some possibilities to help to improve the doctor-patient relationship from the point of media coverage. In this study, we collect 90,499 news articles in total from online mainstream media, such as People's Daily, Tencent News, NetEase News, etc. First, we analyze the news by extracting keywords using three algorithms as word frequency, TF-IDF and TextRank respectively. The results are compared and TextRank is shown the most effective. Second, we analyze the news by building the Latent Dirichlet Allocation topic model and extract news topics automatically. And then these topics are grouped into five themes. Finally, we discuss the characteristics of hospital violence news according to the keywords and topics.