Comparison of identification methods for urban functional areas based on point of interest data
As the carrier of social and economic activities,urban functional areas are of great significance to urban resource allocation and planning management.The traditional identification method for urban functional areas has the defects of strong subjectivity,low efficiency,and poor accuracy.In view of this,this paper introduced the latent Dirichlet allocation(LDA)model in natural language processing(NLP)and the term frequency-inverse document frequency(TF-IDF)model to explore semantic information of urban point of interest(POI)data and reveal regional potential function utilization patterns.Firstly,the urban space was divided into a 500 m×500 m granular grid,and the POI data was mapped to the corresponding geographical grid units.The corpus was built based on the bag of word model.Secondly,the LDA model and TF-IDF model were used to calculate the distribution patterns between grid units and POI data,so as to identify urban functional areas.Finally,the identification results of urban functional areas were compared with the Baidu electronic map and street view images to evaluate the accuracy.The experimental results show that the accuracy of the LDA model is 78%,which is higher than 63%of the TF-IDF model.The LDA algorithm can identify the function utilization patterns of urban functional areas more accurately and can achieve better identification effects in the functional areas that are hard to distinguish by the TF-IDF model.This paper reveals the potential semantic relationship between POI data and urban functional areas,which can be used as a reference and supplement for the research on urban functional areas.It can also assist urban planners in dynamically monitoring urban structure and guide the layout of future urban renewal and development.
urban functional areapoint of interest(POI)natural language processing(NLP)term frequency-inverse document frequency(TF-IDF)modellatent Dirichlet allocation(LDA)
万方数据
维普
