| With the popularization and development of the Internet,the amount of information on the Internet is increasing at an explosive rate.People are also accustomed to browsing and getting information through the Internet,and online news is an important way of network information transmission.Online news site to provide users with news browsing,often also provide users with a comment channel,users can post comments on the news through the Internet.News comment has also become an important part of the news.Web crawler technology can be targeted for the collection of data needed quickly and efficiently to obtain the required news and comment data in the mass data.The rise and application of machine learning provide theoretical support for better predictive analysis in the fields of data analysis and data mining.At present,there are relatively few studies on news and comment compared to Weibo and forums,and lack of quantitative analysis and applied research.The dissemination of news and online reviews is often associated with hot social events,social media trends and so on.In this paper,we pay attention to the use of network crawler technology for collecting network news and its corresponding comment;the distribution and the law of occurrence of research news and comment.On this basis,we adopt machine learning technology to predict news heat,which can provide theoretical support to collection of online news comments,analysis of hot issues and the concern-specific behaviors characteristics of general public.The main contents of this paper are as follows:(1)Through the research of general web crawler technology and the timing characteristics of news comments,we propose and implement the efficient web crawler system for web news and comment of specific target web site according to the demand of collecting web news and its comments.(2)Study the online news comments' spatial distribution among online news and temporal distribution during their appearance.And then based on the news content,analyze and compare the distribution characteristics of different types of news and comments;finally,according to the behavior characteristics of users' comment,define the hot news according to the number of comments owned by the news and in-depthly study the formation and distribution of hot news.(3)According to the research on the distribution characteristics of news and comments,the method of predicting the popularity of news are studied by using a variety of machine learning classification algorithms and some feature information of the sequence of news comments and time of news release.On the basis of this,a feature selection method combining multiple characteristics of indicators and an ensemble learning algorithm that combines multiple ensemble methods are proposed,which could improve predictive performance. |
PDF Full Text Request