| Vietnam has a close relationship with China,and it is very important to analyze the evolution of the topic from time to time in the collection of textual topics in the Chinese and Vietnamese news topics,which is of great significance to the cultural exchanges between the two peoples.The topic evolution analysis technique is intended to present the topic of interest to the user in a concise and orderly manner,which can help the user clearly understand the whole context of the topic.The Chinese topic text collection is a collection of texts that describe the same content in two languages,and because of the language in the text,it contains elements of meaning or similar events,such as objects,time,place,and event triggers.Using this feature in the set of textbooks,we can construct the topic elements of Chinese and Vietnamese to connect the two languages together.In this paper,we focus on the existing textbook collection of Chinese and Vietnamese topics,use the evolutionary analysis method based on sub-topic association,and complete the following two characteristics:1.Proposes a hypertext-based Chinese-Vietnamese bilingual news topic element extraction method.Firstly,according to the method of triggering word excitation,the event elements in the news are extracted,and then the topic hypergraph model is constructed.The sentence of the Chinese-Vietnamese event element is used as the node,the sentence in the text of the Chinese-Vietnamese text is used as the super-edge,according to the probability evaluation function of computing the initial weight of the node and the super-edge is calculated.Finally,the PageRank random walk method is used to score the event elements,resulting in more Chinese-Vietnamese topic elements.The experimental results show that the method has a significant effect compared with the single factor extraction method.2.An evolutionary analysis method of Chinese and Vietnamese bilingual topics based on sub-topic association is proposed.Firstly,the k-means algorithm is used to get the initial sub-topic set,and the initial sub-topic set that has been obtained is taken as the sample instance.The sub-topic collection of each time slice is obtained by single-pass clustering algorithm based on knn algorithm.And then use the cosine formula and KL distance of the mixed formula to calculate the different time window within the sub-topic similarity value.Finally,through the topic evolution analysis step proposed in this paper,we get the relationship of sub-topic between different time slices.The method proposed in this paper is more effective than the method of calculating only the KL distance or the cosine formula. |
PDF Full Text Request