Design And Implementation Of Low Quality Audit System For Article Content

With the maturity of mobile Internet technology,network self-Media platform ushered in an upsurge of development.With the increasing demand for information and information,the network self-Media platform has diversified from simply publishing news and information to providing life,entertainment,society,finance and other aspects.Compared with traditional media,network self-Media platform has higher autonomy and relatively loose speech scale.At the same time,the entry threshold of self-Media platform is low,which leads to the uneven quality of articles.If the article can not be effectively screened,it will probably lead to the wrong direction of public opinion.To review the content of published articles and exclude articles containing low-quality information,so as to ensure a positive reading environment is the key problem faced by various network self-Media platforms.However,in the era of information explosion,millions of articles are produced every day.Obviously,the traditional manual auditing method not only produces higher human resources costs,but also can not guarantee the timeliness of information and achieve a better effect.Therefore,the key method to solve the above problems is to use the means of machine auditing to detect articles.The article content low quality audit system designed in this paper uses machine learning and deep learning algorithms to audit articles.Starting from the text content and image content of the article,to check whether the article meets the requirements.The author independently designed and completed the following three modules:(1)The politically sensitive auditing module provides the identification of the politically sensitive degree of the article.For example,articles containing descriptions that endanger national social security and stability will be identified as politically sensitive articles.Political sensitive audit module is identified by machine learning and deep learning algorithm model.(2)The vulgar pornography auditing module provides the function of recognizing vulgar pornography of articles.According to the recognition of pornographic degree of the pictures attached to the title,content and article,a prediction of the total vulgar pornographic degree is obtained by synthesizing the recognition results.(3)Advertising marketing audit module provides the identification of whether an article is an advertising marketing text.In the process of recognition,the text content of the article will be identified as advertising marketing text,and the accompanying pictures of the article will be checked whether there is a matrix two-dimensional code(QR Code)for advertising.The article content low quality auditing system is applied to the auditing of articles published by network self-media.The online test results show that after the application of the low-quality review system,the number of all kinds of illegal articles is significantly reduced compared with that before the system goes online,which improves the reading environment and achieves the design purpose of the system.