Microblog Hot Topic Analysis And Prediction System

With the rapid development of the Internet,the number of network infornationization participants is increasing year by year.Social communication is not limited to the single form of oral communication between relatives and friends.People's access to information is not limited to traditional media such as newspaper access or advertising.People prefer to take out personal PCs and mobile phones to search for relevant information and comment on the Internet before going out.Sina Weibo has become the first choice for personal users because of its wide coverage and good user experience.And how to make your topics and products spread like viruses is probably the dream of every content maker and entrepreneur.Although the existing microblog public opinion analysis platform in the market can satisfy some users'analysis and visualization of basic features of microblog,most platforms can not provide retrospective download of topic data and analysis and prediction of topic propagation cycle.This paper redesigns the existing hot topic analysis system according to the actual needs of promotion work of a 4A advertising company.This system is a hot topic analysis system based on python.It has the functions of hot topic recommendation,hot topic analysis,hot topic classification,and hot topic trend prediction.At the same time,based on the starting point of crowdsourcing wisdom,it establishes an analyst online analysis platform to meet the higher analysis requirements of customers.This paper mainly includes the following aspects:1.A set of scientific data collection and processing system is designed.The system can collect topic data in real time and d:irectionally.It provides keyword search,time slice search and topic cycle information collection.It also stores topic-related user information and microblog information.It has the ability to collect related information horizontally.At the same time,in addition to the general data cleaning work,the system combines the frontier direction of science,identifies the false advertising infornation collected by the data,and marks statistics,which provides more reliable data input for model training in the system.2.According to the actual business needs of users,based on the principle of user-friendliness,users can reasonably recommend and build a data warehouse to provide users with data backtracking download.Considering the needs of platform users for topic promotion,the platform provides similar historical topics for platform users,relevant active user recommendation,and official promotion platform promotion functions.Visual display of user analysis data is also carried out.3.According to the observation of the life cycle of historical data,it is found that the life cycle consists of two parts:preheating period and active period.The system classifies and predicts the two parts of the life cycle based on the deep learning convolutional neural network(CNN)respectively.It ensures that the analysis of topic data is the prediction of hot topics in the active period,and can significantly improve the accuracy of topic prediction.Based on the neural network,the system can predict the hot topics in the active period.Nonlinear regression is used to predict the active period.4.Based on the principle of standardization of platform system development and convenient maintenance,the platform builds a micro-blog hot topic analysis and prediction system based on Web services.The system adopts Django architecture,which layers the system horizontally,extends deep technology,reduces interlayer coupling,increases intra-layer aggregation capability,and facilitates development and maintenance.5.In order to satisfy the personalized requirement of platform users for hot topic public opinion analysis,the system innovatively puts forward the platform analyst role according to the principle of public wisdom creation.Platform analysts can view the user's release of relevant analysis requests in the analyst interface,download the relevant analysis data base and submit individual analysis reports according to the user's personalized needs.Make the system function more perfect,user friendly.