Research On The Dynamic Evolution Trend Of Topic Popularity In Online Health Communities Based On The LDA Model

In the context of online health communities,analyzing user’s posting content and studying the dynamic evolution of the topic’s popularity may help to understand how user’s health issues and needs evolve over time,which can facilitate online health communities provide better online health services.A particular diabetes community in China(Tianmijiayuan)was chosen as the research context.Users’ posting data in type 2 diabetes section was crawled from this community.A LDA model was used to identify themes and extract keywords,and these themes were grouped into 7 main topics based on keywords.For each topic,this paper used kernel density estimation to investigate the distribution pattern of topic popularity.Finally,this paper uses the growth model to study the growth trend of themes’ posts in each topic to analyze the reasons for the dynamic evolution of user attention over time.Based on LDA model,seven main topics were identified in Tianmijiayuan.These topics are: diagnosis and inspection,medicine,treatment,control,diet and exercise,emotion,and social interaction.The results indicate that users’ attention in this community demonstrate different patterns for different topics.Concerning control and medicine topics,users’ attention varies with time.Concerning topics like emotion and treatment,users’ attention declines with time.Concerning topics like diagnosis and inspection,diet and exercise,and social interaction,users’ attention increases with time.The research on the growth models of themes’ posts shows that the difference in the degree of attention of different topics is closely related to the proportion of growth modes in the topic.