Filtering And Sorting Weibo Hot Comments

With the rapid development of the Internet,the Internet has achieved a leap from Web1.0to Web2.0,and information dissemination has also changed from a single method to an interactive sharing method.Among them,Weibo platform has been widely promoted.Users can get current hot information through Weibo,and also can share their daily lives or hobbies by publishing Weibo.Weibo,a platform that can be shared interactively,has gradually become an essential part of people's lives.However,some users or manufacturers will make some irrelevant comments due to certain emotions or certain interests.These comments seriously affect the user's mood and even mislead the user.Secondly,these comments also waste platform resources and pollute the platform environment.Therefore,filtering and ranking Weibo comments is a meaningful task.The purpose of this article is to effectively filter Weibo comments by establishing a model.The main process is: determining a Weibo topic,crawling comments under a specified Weibo topic,randomly extracting certain sample data,vectorizing the comments,and using Logistic regression model,SVM model,Knn model,Bert model and combination model are used to complete the screening of reviews.And use the filtered reviews to mine valid opinions from the reviews,formulate a formula for calculating the review quality score,and sort the reviews.The innovations of this article are: 1.On the problem-cutting side,because the usual comments are sorted according to the number of likes,there may be many effective comments that are behind and cannot be seen by users.Let users see more effective reviews.2.In the selection of the model,the Bert model was released by Google in 2018.This model created the best results in 11 different NLP tests.By combining the bert model and other base classification models,the model's The classification model is more accurate.