Construction And Application Of Domain Think Tank Based On Wechat Official Accounts

With the exponential growth of internet information resources,people have entered the information age with extremely rich information resources.It is difficult for people to quickly get the information they want from a vast amount of information resources.Therefore,how to build a Think Tank of information integration to provide users with more valuable information resources has gradually become a research hotspot.To alleviate the problem of disorganized information on the internet and facilitate users to obtain the required information more accurately,this thesis takes WeChat official accounts,an entry-level application for internet information,as the research object to study how to rank the WeChat official accounts and classify the WeChat articles.A method for ranking the WeChat official accounts based on LambdaMART algorithm and a method for classifying the WeChat articles for information services are proposed.Firstly,this thesis proposes a method based on LambdaMART algorithm for ranking the WeChat official accounts by investigating the current ranking methods and considering the characteristics of the WeChat official accounts.Most of the existing methods rank WeChat official accounts based on the quantitative indicators such as reading numbers and praising numbers and weighted through human experience,which neglect the influence of the article content.To solve this problem,on the basis of retaining the quantitative indicators,we propose four ranking features based on WeChat article content,which are theme verticality,text posting stability,topic coverage and topic relevance.LambdaMART algorithm is used to training model to rank based on all the features,and feature selection is optimized by principal component analysis.Then,a method for extracting features based on article genre is proposed to classify WeChat articles for information services,and compared with the existing methods.Finally,the domain Think Tank construction and application system based on WeChat official accounts is designed and implemented,which includes a module for ranking WeChat official accounts and a module for classifying the WeChat articles.Experimental results show that:(1)The method based on LambdaMART algorithm for ranking the WeChat official accounts has obtained better ranking results than other methods,and the effectiveness of the four ranking features proposed in this thesis has been proved.(2)Compared with the existing methods,the method of extracting features based on article genre has obtained better results for classifying the WeChat articles for information services.