Association Analysis Of Chinese-Vietnamese Bilingual News Events

With the implementation of the national strategy of the Belt and Road Initiative,the exchanges and cooperation between China and Vietnam are getting closer.Keeping abreast of the dynamics of news events in both countries is important.Reports on the Internet by Chinese and Vietnamese media provides comprehensive information for understanding of relevant events in both countries.This paper studies the method of correlative analysis of bilingual news events.Designed to make use of the internet news on the Internet and discover hot issues of common interest in China and Vietnam and the links between these eventsThrough the bilingual news to find events that are of common concern to China and Vietnam,the key issues faced in events correlation analysis are:1.There is currently no public dataset for training and evaluation of bilingual news event association analysis.We lack data for model training and evaluation.2.Both Chinese and Vietnamese media have different focuses and attitudes when reporting on the same incident.How to cluster bilingual news reporting the same event is a big challenge.3.News events are not isolation.How to calculate the influence of news events is a big challenge.This paper studies the construction of Chinese-Vietnamese bilingual event correlation analysis datasets for these key issues,the Chinese-Vietnamese bilingual news event clustering method,and the bilingual event correlation analysis method,and achieved the following results:(1)Constructed a Chinese-Vietnam bilingual event correlation analysis data set.Building Small-Scale Chinese-Vietnamese Aligned Corpus.The Chinese-Vietnamese double-sentence aligned corpus is used to construct bilingual vector space,and the bilingual news is uniformly represented in the same feature space.Twenty event clusters were constructed to evaluate the classification effect of bilingual news events in Han and Vietnam.A total of 600 related event news collections and 600 unrelated event news collections were constructed to evaluate the effectiveness of cross-language news event association analysis methods.(2)Propose the method of clustering Chinese-Vietnamese bilingual news events.This paper first uses the Chinese-Vietnamese aligned corpus to construct a Chinese-Vietnamese bilingual vector space based on word meaning,and puts Chinese and Vietnamese bilingual news in the same feature space.As for the characteristics of news events,this article uses density-based news clustering methods and event elements to clustering news.News that report the same event are clustered into the same cluster.The experimental results show that this method effectively improves the effect of cross-language news event categorization.(3)Propose a local intimacy propagation algorithm based on factor graph.First,we use bilingual topic model to get the bilingual topics and topic probabilistic distributions from bilingual document.Then we built events' factor graph based on event text similarity.Using local intimacy propagation algorithm to compute influence the for interrelated events on the factor graph under the same topic.Finally we got the influence topology of events under different topics.Experiments results show that the method we propose have achieved better effect compared to the traditional method.(4)Using JavaEE to design and implement a prototype system for bilingual event clustering and correlation analysis.Through this system,users can view news on the Internet in China and Vietnam;view events of in China and Vietnam;and news on these events;and view the correlation between bilingual news events.