A classification method of Vietnamese news events based on maximum entropy model
Zhu Li-juan, Zhou Feng, Pan Qing-qing, Yan Xin, Yu Zhengtao · 2015
Based on the characteristics of Vietnamese news texts and selection of Vietnamese words and phrases, part of speech, named entities, news titles, keywords and so on, this paper proposes a classification method of Vietnamese news events based on the maximum entropy model. This method selects the named entities and news keywords in Vietnamese news titles and the event trigger words in the key sentences corresponding to the keywords as the news classification features, and adopts the maximum entropy model to achieve classification. Furthermore, this paper collects more than 6000 Vietnamese news texts, which are marked into seven kinds of news event corpora such as politics, economy and culture, etc. and subject to training, and then obtains classification model of Vietnamese news texts and achieves type classification of Vietnamese news events. The experimental results show that the accuracy of the classification method of Vietnamese news events proposed in this paper reaches 96.97%.