Temporal Relationship Recognition of Chinese and Vietnamese Bilingual News Events Based on BLCATT

Abstract

The temporal relationship identification of bilingual news events is essentially to find the temporal relationship between news events in different languages under the same topic. At present, the event temporal relationship recognition requires a lot of manpower to design a timeline-based template. The implicit semantic information in the sentence is difficult to obtain, and different language texts are difficult to represent in the same feature space. Therefore, it is difficult to obtain the temporal relationship of cross-language news events. To this end, this paper proposes a Bi-LSTM Cross Attention (BLCATT) model to identify the temporal relationship between two events. Firstly, the bilingual word vector is used to represent the Chinese and Vietnamese bilingual news texts. Secondly, using Bi-LSTM to capture the semantic information of the sentence. Using the attention mechanism combined with the trigger word to obtain the event semantic information of the enhanced trigger word information. The event semantic information including the temporal logic relationship is obtained by using the cross-attention mechanism combined with the trigger word and combines the two parts of the event semantic information as the event coding. Finally, the event coding and the event rule feature are combined to obtain the temporal relationship information, and then the event temporal relationship recognition is transformed into the multi-class problem. To solve. The experimental results show that the proposed model achieves good results.