Design And Implementation Of Emotion Analysis Subsystem Of User Portrait Platform For Autohome

Because of the development of the Internet,the impact of online social and online shopping on all aspects of people's lives is gradually increasing.Although most people won't buy cars through the Internet,it doesn't affect their understanding of the intended car series through the Internet,as driving experience,fuel consumption,cost performance,etc.At present,the autohome platform has given scores on fuel consumption,interior decoration,cost performance and other aspects of each car series,but further analysis of the crowd portrait is still missing.In this thesis,the more mature text emotional analysis method is applied to the analysis of word-of-mouth data,from which the multiple feedback information of the car buyers to each car series is obtained.On one hand,an emotional dictionary in the vertical field of automobile industry is constructed to determine the emotional polarity of this group through the part of speech tagging.On the other hand,the random forest algorithm is used to analyze the emotional polarity of comments.In addition,the eigenvalues obtained from emotion dictionary analysis are applied to the classification of random forest algorithm,that is,the combination of emotion dictionary and random forest is used for analysis.According to the actual business needs,the analysis of this thesis takes the car series as the unit,in order to obtain consumers' satisfaction with the car series.This can not only help the subsequent consumers to understand the relevant information of the intended vehicle series,but also has a certain reference significance for the automobile manufacturers to carry out the iterative optimization of the automobile.The main work completed by the author is as follows:(1)The emotional analysis module is built in this thesis,and the feasibility requirements,functional requirements,non functional requirements of the module,and design the overall analysis framework is analyzed in this thesis.(2)It realizes the functions of the core function module,the core algorithm module and the emotion analysis module.In the core algorithm module,the algorithm with good experimental results is used as the core algorithm,and the data analysis is realized in the emotion analysis module.(3)The results of emotional analysis is verified in this thesis,and the function and performance of the emotional analysis module to verify the availability of the module is tested in this thesis.The results show that: The establishment of emotion analysis subsystem further improves the efficiency of operation and analysts.Under the premise of ensuring the accuracy rate and recall rate,ensuring that the follow-up work can continue,and taking time and labor cost into account,it is more effective to use random forest for analysis.The emotion analysis subsystem implemented by random forest algorithm has better applicability.