Hot Spot News Discovery And System Method Research Based On Data Mining

Internet news has become an important source of public access to information.New network resources and network news applications is increasing,the number of network news shows explosive growth,that brings the user a lot of difficulty to read the news.Discovery and analysis of hot events become important issues need to be resolved from a large number of network news.Although machine learning,natural language processing,and many other techniques have been widely used in network hot events discovery,but the existing text represent model has relative limitations exist,user is still not satisfied by the performance of text representation,as well as many issues need further study.In order to achieve the purpose of a deeper understanding of the text,this article build a cluster-based internet hot event finding method based on chinese sentence structure model.Firstly,this method analysis the document inorde to get sentence meaning element,calculate word weight to generate semantic vectors;use semantic vector to hot event finding system,use clustering algorithm that combine single-pass clustering ideas,coherency hierarchical clustering and K-means clustering algorithm,the accuracy is 75.2%.In addition,build a method of simplified representation of the event,extract the key points of events and event tags,key points of events accuracy is58.9%.In addition,design and achieve a prototype system of hot events finding and event simplified representation.