| With the development of the Internet, reading network news has become a better choice for people who are care about the current events. Network news has significant characteristics: large quantity, various, and report views are not the same. The users’ desire is to read interested network news quickly, not only can save time, but also can improve the reading experience. Although there have been a variety of news platform, but they don’t handle the news more precisely, especially in the field of education, usually few topics become hot spots. So I design and implement education news platform for students, parents, educators, by capturing education news from the Internet, with the steps of processing, mining, providing services, such as news retrieval and hot education news topics recommendation.In this paper, I design the kapok education news platform and divide it into several modules, include web crawler, information extraction, texts duplication, news indexing, classifier train and hot topics detection module. Web Crawler is for crawling web pages, information extraction module is for extracting key information from the original pages, texts duplication module is to identify the pages which has been reprinted by many websites, indexer is for indexing news and classifier is trained to classify news, topic detection module is used for detecting education hot topics.In this paper, education news classification and hot news topics detection are treated as two key issues. For a more fine news classification, we designed a hierarchical classification method for education news. According to the characteristics of education news, adjusting the feature weight and feature proportion are done to improve classification result. To identify the focus of education news, topic detection and tracking and hotspot recognition should be done. In the first step, agglomerative hierarchical clustering and Single-Pass method are combined to detect topics, besides, nouns are selected as clustering features. In the second step, the purity of cluster, the number of news text and media number are used to calculate clusters’ hotness. Experiments have been done to demonstrate the effectiveness of the method.In this paper, I describe the mechanism of kapok education news platform in detail. The system can update data incrementally, only process the lately crawled news each time and has no impact on the indexed news.The running and good performance of kapok education news platform demonstrate the correctness of the design method and the good system implementation. |
PDF Full Text Request