An Method of Improved HLDA-Based Multi-document Automatic Summarization of Chinese News

There are a lot of Chinese news about the same topic on the Internet today. Many of them are similar or repetitive for readers. It is hard to find what are the readers needed exactly. Multi-document news summarization aim at extractioninformationfrommultiple news texts on sametopie to automatically generate summary report for readers. Our paper chooses the news of the Great Wall as an example to illustrate the method of automatic summary generation In ourmethod, combinedwiththe characteristies ofnews corpus, the HLDA topie importance calculation model is improved. Based on the abstractly characteristics of the model, news related features such as news headline words, topie sensitive words and TF-IDF are added. Abstract sentence extraction and sentence fusion, automatic generation of abstracts. Experimental results show that the proposed algorithm is higherin the index thanthe traditional method, indicatingthe accuracy of the corpus combined with news features and the improved HLDA algorithm.