Abstract:Objective To obtain the comment data of the on-site service platform using the web crawler, and to explore the users′ focus and emotional attitude on the on-site service using text mining technology. Methods Python code was used to crawl the comment data of door-to-door nursing service platform, latent dirichlet allocation theme model mining was utilized to obtain the potential theme of comment, and ROSTCM6.0 software was used to calculate high-frequency term and analyze emotion theme, and mine the hotspots, core themes, and comment emotion tendencies of nurses and patients. Results A total of 3 166 comments were captured, and the topic model identified 6 potential topics, namely platform function, service price and efficiency, service content and professionalism, nurse attitude and patient experience, service objects, and platform download and use.Words such as nurse, ser-vice, attitude, convenience and platform were highly mentioned by users, which showed the users′ feelings and focus to some extent; users′ positive emotions accounted for 59.19%, neutral emotions took up 27.23%, and negative emotions accounted for 13.58%, and users′ negative emotions mainly focused on differences in service levels, unreasonable charges, and platform download and use, etc. Conclusion Users have a good perception of the overall service of each platform, but there are still dilemmas such as uneven service levels, irregular charges, and irregular platforms.It is recommended to improve accurately for the shortcomings in the development process, so as to promote the healthy development of door-to-door nursing services.