Research And Construction Of Network Hot News System Based On Text Mining Technology

In view of the complex and diverse characteristics of news content and the problem that the current traditional news search cannot completely control the user's demand results,the main work of this article is to study the effects and laws of commonly used news text mining algorithms and text search recall algorithms.The characteristics of the user's search terms and news text in the search behavior,explore the definition and abstraction of the user's search terms and news text information into a characteristic form,and transform the traditional news search into a dialogue mode between the user and the news robot,Provide support for further improving the accuracy and satisfaction of users' search resultsIn order to realize the intelligent news search in the form of dialogue,first of all,select the appropriate news data source and write a web crawler to crawl the news.This article aims to provide users with hot news,so the crawler crawls the website's daily hot list news.This article compares four mainstream news websites,and finally selects Sina news website to provide data for the system.Then,combining BERT and BiLSTM-CRF as the construction method of named entity recognition model,and using BERT to optimize the process of feature extraction and vectorization of TextRank,to achieve the extraction of news topics and abstracts.When a user conducts an intelligent conversation,the named entity recognition model recognizes the user's chat content,and returns the news topic,abstract,and original link information corresponding to the identified entity.The user can quickly understand the news content through the topic and summary,and browse through the original link for detailed browsing news content.This system has significantly improved the convergence,accuracy,and recall rate of model training,especially in capturing important information in sentences.Smart chat combines the characteristics of users' search terms with the core information of news.Features are matched to obtain more intelligent news information results that can better integrate with user intentionsIn order to realize a complete news platform system,this article builds a news system and a news management system through webpage development technology.The news system implements interactive functions such as user intelligent news chat,browsing hot news,news search,and comment likes.News management system It provides user management,news management,news review,user comment management and other functions for administrator users.The news platform system in this article realizes the complete operation and management functions of the news website.