Abstract: Taking 30 public libraries in China as the research object，the content of the public comment network in the past five years as a sample，using the content analysis，extracting corresponding high-frequency vocabularies and constructing a semantic web to analyze the public library image.It is found that users take a positive attitude on the overall image of public libraries，and they tend to vent bad moods on the network platform，which shows the shortcomings of the image of public library.The public library image can be optimized from four aspects：paying attention to the environment of public library and the construction of collection resources，improving the comprehensive quality of public library staff，holding colorful activities and focusing on children's reading service.

Key words: public library；dianping.com；image