AbstractIn rapidly developing large cities, land functions change rapidly and are mixed. It is essential to obtain accurate functional area maps to evaluate the urbanization process and its impacts on rational urban planning. Many studies have verified the usability of point of interest (POI) data in identifying functional areas. However, few studies have considered the impact of the size differences of the entity objects represented by POIs on the functional categories. A new scoring evaluation model that combines the area score and the normalized kernel density value was constructed to identify urban functional areas. The area score was obtained based on the building vector data and the constructed ambiguity function. The kernel density value was obtained by kernel density analysis of various POIs. The two factors were combined to calculate the influencing scores. Functional areas were identified by the proportion of the influencing scores. Compared with the traditional model based on the quantity of POIs and tested in the system of different functional zones, the scoring evaluation model improved the overall accuracy by 17.4%, and the kappa coefficient increased from 0.51 to 0.73, with strong robustness. Therefore, the model constructed can provide an accurate full-coverage regional functional area map, which can help planners obtain urban spatial structures and make scientific decisions.Practical ApplicationsBased on POI data, a scoring evaluation model was constructed to identify urban functional areas (commercial zones, public service zones, residential zones, industrial zones, transportation zones, and scenic zones). The principle of the model was to multiply the area score of each POI by the normalized kernel density value to obtain the impact score of each point and to identify functional areas by calculating the proportion of the impact score of various functional POIs in each plot. The area score was obtained based on the building vector data and the constructed ambiguity function, and the kernel density were obtained by the kernel density analysis of various POI data. The overall recognition accuracy of this model was 83.7%, which is suitable for different functional area systems and research areas. In urban research and planning, this model can be used to quickly obtain information on functional areas based on POI data to master the spatial pattern of the region and analyze the functional structure in depth to make scientific decisions.