Analyzing the complex spatial structure of the city, identifying the distribution characteristics of urban functional facilities and the types of regional urban functional zones can provide important data support for optimizing the spatial layout of the city and promoting the sustainable development of the city. A city functional area recognition method based on TF-IDF (Term Frequency-Inverse Document Frequency) algorithm is proposed, which integrates OSM (Open Street Map) road network and POI (Point of Interest) data, taking the urban area of Hefei city as the research object. Firstly, we analyze the spatial distribution pattern of POI by using average nearest neighbor and hotspot analysis; then we use OSM road network to divide the study area, obtain the basic study units which are independent of each other and have socio-economic functions, and then use TF-IDF algorithm to weight the POI data, and at the same time, combine with the frequency density ratio to identify the division of the city's functional areas, and then compare and analyze the spatial distribution characteristics of the different functional areas. The results show that all types of POIs show obvious clustering distribution characteristics, and the spatial clustering patterns of cold and hot spots in the three types of POIs, namely public, residential, and commercial services, are relatively similar, but the degree of clustering is higher in commercial land. Urban functional zones include twenty-one types of functional zones, green space and traffic-related functional zones are concentrated in local key areas, the spatial distribution patterns of public, commercial and residential are relatively close, and the distribution of industrial functional zones is very consistent with the industrial development zones of Hefei, so that the overall results of the identification are in line with the actual situation of Hefei.