Design And Implementation Of Abnormal User Detection Module Based On Public Opinion Analysis System

Social network has become an indispensable part of people’s life,but the gathering of negative public opinion in social network leads to serious public opinion crisis and serious harm to social harmony and stability.Through public opinion analysis technology,especially abnormal user detection technology,we can timely find the negative public opinion in the online community and the users who publish negative public opinion,and monitor the online public opinion,which can provide decision support for the public opinion governance of relevant departments and effectively prevent the outbreak of public opinion crisis.Aiming at the lack of domain oriented sensitive knowledge base,a domain oriented sensitive knowledge base is established by extracting the characteristics of social network sensitive content.A word based adaptive entity recognition model is proposed.Based on this model,people,places and events in historical sensitive data are extracted,and a domain oriented sensitive event database is constructed.Based on the mainstream emotion dictionary,a domain oriented emotion thesaurus is constructed with reference to sentiwordnet polarity annotation,and the emotion score of words in the thesaurus is taken as the weight parameter to measure the sensitivity.Aiming at the coarse-grained problem of user classification in the mainstream user identification model,an abnormal user detection model based on sensitive computing is proposed.According to the characteristics of typical users,domain oriented sensitive user types are designed.Through the sensitive knowledge base and sensitive calculation,the sensitive type and sensitive score of the sentence are obtained,the sensitive users are identified according to the sensitivity discrimination results of the text,and the user sensitivity tracking analysis is realized based on the temporal characteristics.In view of the problems of single dimension,lack of accuracy and realtime in the mainstream public opinion analysis system,a domain oriented public opinion analysis system is designed and implemented.The core functions and performance indicators of the system are obtained from the demand analysis,the architecture,functional modules and database of the system are briefly designed,and the functional modules of the system are designed and realized in detail.The system follows the principle of high cohesion and low coupling,adopts the front and rear end separation mode,and uses SpringBoot+Vue.js technology stack realizes the network public opinion monitoring with comprehensive collection,efficient response,intelligence and reliability.This paper carries out the function test and performance test of the public opinion analysis system from the perspective of software engineering.The results show that the system function test results are consistent with expectations,and the performance test is good.The effectiveness of the abnormal user detection model based on sensitive computing is verified.The F value of the abnormal user detection model is higher than that of the comparative experiment,which meets the needs of the public opinion analysis system.The system has been deployed and operated in script based on docker technology.It is applied in the field of education and runs well.