Public Opinion Analysis Based On Textual Orientation

Network public opinion analysis aims at the Internet platform for the collection of public opinion information,processing,analysis and warning.Contrast to the traditional media,network public opinion can truly reflect the feelings of the people's feelings.Especially for the special events,netizens know about the development of events through various channels and express personal views.Thus,collecting network public opinion and mining the information can effectively understand the changes in emotional tendencies of Internet users.We can response to the situation in time and guide the development direction of public opinion.This work analyzes the steps of network public opinion analysis,according to the characteristics of the network public opinion information data,this paper proposes a word segmentation method based on the manually adjusted mixed word segmentation model and calculates the weight of emotional words with the feature extraction technique.Based on classification algorithm,we construct the text emotion tendency classifier and use the classifier to judge the emotional tendencies of the data.Taking the " Samsung Note7” event as an example,we have done some research about network public opinion analysis.The Study found that:First,compared to the traditional single word segmentation method,the artificial adjustment of the mixed word segmentation model has improved effectively in the text of the word segmentation.Second,the w-svm is better to judge the text tendency of the event text data.As the classification algorithm evaluation index,AUC index upgrade from 0.7 to 0.76.Third,when facing the unexpected events,well-known enterprises(such as Samsung)should analyze the network public opinion information in time and get the emotional tendencies of netizens.When the public opinion environment in a controlled state,they should response to the actual needs of netizens positively instead of avoiding their responsibility.