Viewpoint Recognition And Sentiment Orientation Prediction For Public Opinion In Urban Planning

With the development of information technology,the Internet has become an important channel for the public to express their opinions.With the construction of smart city,the network public opinion in the field of government affairs has been widely concerned.In the field of urban planning,the public can get information of urban planning cases in the planning stage through the government web site and at the same time express their attitude,opinions,satisfaction,etc.named feedbacks,which is a large number of micro individual data for urban planning public opinion analysis.Therefore,the feedbacks of urban planning cases is an important way to obtain the public opinion information,which has the characteristics of fine granularity,pertinence and timing.Based on the urban planning feedback data,it is of great significance to realize the point of view recognition of urban planning public opinion and to predict the public opinion characteristics of urban planning cases in the future such as opinion category,emotional tendency and attention people groups according to the public opinion of similar history urban planning cases.This paper presents a method for forecasting the future public opinion of urban planning cases.The description and expression model of urban planning cases is constructed based on the geography case theory.Based on the text mining method,the multi-granularity public opinion features of urban planning cases are extracted,and the quantitative expression of case results is realized.The quantitative expression of case attributes is realized by combining urban planning land type,POI data and population census data.According to the case attribute expression model,the measurement of case similarity was achieved,and the prediction model of public sentiment tendency based on urban planning case-based reasoning was constructed.According to the public sentiment characteristics of historical similar planning cases,the affective tendency of new planning cases in different viewpoints and the concerned groups of the cases were predicted.The main results are as follows:(1)A multi-granularity semantic feature extraction method for public opinion in urban planning field.On the topic-based hierarchical clustering method,a multi-level public opinion classification system in the field of urban planning is constructed,and the opinion and emotional tendency categories annotation of public opinion data is realized by syntax analysis.Based on the user’s dictionary and part of speech annotation,the attention groups of urban planning cases were obtained.The Linear SVM model was used to realize the classification of emotional tendency of feedback text data and extract coarse-grained public opinion features.Point of view recognition and finegrained public opinion features extractions of planning cases are achieved based on the multi-label classification of feedback text.The results show that the classification accuracy of opinions and affective tendencies can reach more than 90%.(2)Quantitative representation of spatial semantics of urban planning cases based on representation learning.First,the planning category keyword dictionary is constructed according to the urban planning landuse type.And the keyword of urban planning case name is extracted based on TFIDF.Then the planning category classification is realized based on keyword matching.The Class2 Vec model was trained by using the urban planning news text data to obtain the category word vector of the urban planning case and realize the semantic feature expression of the case category.The Loc2 Vec model was proposed,and POI data was used to train the model,so as to obtain the semantic feature expression of urban planning location combining spatial distance relation.The results show that the spatial semantic vector can realize the quantification of spatial attributes of cases,and the semantic vector distance can measure the similarity between spatial features of cases,which provides the data basis for the quantitative calculation of case similarity.(3)A new urban planning public opinion prediction model based on the combination of people and land.Based on the census data,the quantitative expression of population attributes of planning cases is realized.According to the case attribute expression model,a case similarity calculation method combining multi-dimensional case attributes is proposed to obtain the historical similarity planning cases and their similarity of the current planning cases.This paper proposes a calculation strategy of emotional tendency based on similarity weight,constructs an sentiment prediction model integrating subjective factors of residents and objective factors of geographical environment,and predicts the multi-granularity public opinion characteristics such as the opinion category of current planning cases,the category of emotional tendency,and the category of people concerned in cases feedbacks.The data of real planning cases are selected for verification,and the results show that the method proposed in this study can realize the prediction of public opinion characteristics of cases,and the prediction accuracy can reach more than 74%.In conclusion,geography case-based reasoning for the urban planning and public opinion prediction provides a new solution,this paper puts forward the urban planning case description and expression of framework to realize effective organization of urban planning case data.This framework on the one hand provides quantitative expression for the extraction of case attributes factors framework,on the other hand provides a basis for the planning case feedback opinion information extraction and the urban planning case similarity calculation.For unstructured text of the urban planning feedback data analysis,this study integrated the text analysis method of the lexical level and textual level to realize the multi-granularity planning case public opinion characteristics extraction and orderly organization.For macro-level public opinion characteristic,the attention groups and emotional tendencies are extracted.For microlevel public opinion,point of view classification system is built and fine-grained emotional tendencies classification are realized.Combining with geography case-based reasoning theory,planning case similarity measurement is achieved based on the multidimensional attributes of the historical similarity planning cases.Public opinion emotional tendency prediction method is built based on the planning similarity weight.The prediction method in this paper has better precision in both public opinion emotional tendency prediction and attention groups prediction,verified the effectiveness and rationality of my model.This research can realize the information extraction of historical planning public opinions and the advance prediction of future planning public opinions,and can provide scientific and reasonable decision support for the planning.