Analysis Method Of Destination Image Perception Based On LDA Model And Dominant Semantic Dimensions

User-generated Contents of tourism,which can directly reflect tourists' behavior pattern,emotional state and other image perception,has become an important data source for tourism destination image research.At present,the research method of image perception mainly relies on the content analysis such as word frequency and word co-occurrence,which lacks the topic of document and cannot find deeper semantic information.The LDA(Latent Dirichlet Allocation)topic model can obtain the topic information of the texts,but it is still very subjective in topic interpretation.In addition,the current tourism image perception mainly focuses on the analysis of text semantic perception,and lacks the study on the spatial and temporal characteristics of it.It is a scientific problem that this paper tries to solve how to further improve the objectivity of the description of tourist image characteristics on the basis of the calculation results of LDA thematic model,and dig into the spatial and temporal pattern of tourist image from different scales.The paper takes the Old Town of Lijiang as an example,and selects this data from microblog published by tourists as the data source.Through LDA model and feature dimension method,the paper describes the image in a semi-quantitative way,and then discovers its spatiotemporal evolution process through kernel density analysis.The advantages of this method are as follows :(1)it can solve the problems faced by LDA in the application of short texts,ensure that different researchers can interpret the basically consistent semantics from the data,and finally obtain the detailed structure and characteristics of image perception.(2)this method can obtain the spatio-temporal evolution process of image perception at various scales by analyzing the influence and constraints of spatial and temporal factors on the perception.The research shows that the method of integrating LDA and feature dimension can extract the topic of the short text of the microblog better,reduce the subjective judgment of image perception to some extent,and get the results basically consistent with the existing research.In addition,the image perception of the Old Town of Lijiang obtained by this method has a more precise structure,realizes the spatialization of image perception,and discovers the spatialtemporal pattern of it.The specific image characteristics are as follows:In the first-level dimension,the basic framework of image perception can be fully outlined,through the 6 groups of core spatial and landscape elements,the special perception of 4 different groups of people,and the interaction characteristics of human and environmental elements;and in the second-level dimension,more detailed perception of destination image can be vividly presented from three aspects: the slow living in the Old Town of Lijiang from 7 perspectives,the culture of nightlife and affairs from3 perspectives,and tourists' perception of the fusion of local culture and modernity based on 3 vectors.The LDA model has successfully depicted the image of the Old Town of Lijiang,and can further analyze the causes and mechanisms of image formation,providing a new perspective and a new method for the study of image perception of tourist destinations.This paper studies the spatial distribution of image perception from two perspectives: feature dimension and spatio-temporal evolution.From the perspective of the images of first-level feature dimensions,the high-density core areas of behavioral and psychological dimension images are more widely distributed.Two distribution patterns are found from the spatial distribution of second-level feature dimensional images: single density core distribution and multiple density core distribution,the spatial distribution of image with specific POI is mainly belong to the former.In terms of the intraday time scale and the tourism season scale,women's perception of each topic is stronger than men's,and that of foreign tourists is stronger than that of local tourists;From the study of intraday time scale,three spatial evolution patterns of the image are found: density core diffusion pattern,density core convergence pattern and density core diffusion and then convergence pattern.