Design And Realization Of The Comment Analysis System Of Police Notification

Under the condition of strict control,text analysis has the advantages of easy access,high value and great effect,and plays a great role in the field of public opinion management and control.In practical applications,complex comment texts are mostly collected,which leads to problems such as low accuracy of sentiment classification and poor effect of topic extraction,so it is difficult to meet the needs of practical applications.Therefore,text analysis technology has become a research hotspot.This paper mainly studies the analysis method of LDA subject model based on improved particle swarm optimization(SVM)and visual analysis,and designs and implements the comment analysis software of alert notification.In the aspect of text sentiment classification,a text sentiment classification method based on improved particle swarm optimization SVM classification model is proposed.In the stage of searching for the optimal parameters,the improved particle swarm optimization algorithm of inertia weight is used as a tool,and the function based on the change rate of particle fitness value is used to dynamically adjust the inertia weight coefficient of particle swarm,so that the global searching ability of particles is provided and the local optimum is not guaranteed.The standard PSO,APSO,AIWPSO and the method proposed in this paper are simulated and tested on four benchmark functions of Sphere,Schewefel,Rastrigin and Rosenbrock respectively.The experimental results show that The improved inertial weight particle swarm optimization algorithm is superior to other algorithms in mean value and standard deviation,and can quickly converge to the optimal value,and has better performance.In the stage of text sentiment classification,the improved inertial weight particle swarm optimization algorithm combined with SVM can find the best parameters for SVM efficiently and input them into the model.The simulation experiments and performance tests of SVM,SVM+PSO and the proposed method are carried out on the comment text data set obtained and processed in advance.The experiments show that the accuracy of SVM sentiment classification method based on improved inertial weight particle swarm optimization is 3.37%and 1.55% higher than that of SVM and SVM+PSO,respectively.The accuracy is improved by 3.69% and 0.78%,respectively,and the recall rate reaches 96.45%.The experimental results show that the proposed method has better performance.In the aspect of subject information extraction of comment text,a method of subject information extraction based on visual display LDA is proposed.First,we calculate the confusion degree of positive and negative comments respectively to determine the best number of topics.Then,the subject model is trained according to the determined number of subjects,and the Pyl Davis library is called to realize the interactive visual display of subject information extraction.Finally,through the analysis of the visual display information,not only can the number of subjects be checked to determine whether it is reasonable,but also the potential topic information can be obtained by adjusting the parameters.The example of topic extraction shows that this method can cluster topics effectively.In terms of the design and implementation of the alert bulletin comment analysis software,Jupyter Lab is taken as the development platform,PyQT5 design interface is used,and Sklean,GenSim,PylDavis and other third-party libraries are combined to design and implement the alert bulletin comment analysis software.It has the functions of limiting user rights,obtaining text information,classifying text emotion,extracting text subject information and interactive visual verification.