Chinese-Vietnamese news event time sequence relationship identification method based on bidirectional cross attention mechanism

Abstract

The invention relates to a Chinese-Vietnamese news event time sequence relation recognition method based on a bidirectional cross attention mechanism, and belongs to the technical field of natural language processing. According to the invention, sentence semantic information is captured by using a bidirectional recurrent neural network; triggering word semantic information in the event semantic information is enhanced through an attention mechanism; time sequence logic semantic information is obtained through the cross attention mechanism, the three parts of semantic information are embedded into event codes, finally, the event codes and inter-event rule features are fused and input into a classification layer, an event time sequence relation is output, and then event time sequence relation recognition is achieved. At present, event time sequence relation recognition needs a large amount of manpower to design a template based on a time axis, implicit semantic information in sentences is difficult to obtain, different language texts are difficult to be represented in the same feature space, and the problem that cross-language news event time sequence relation obtaining is difficultis solved.

Description

technical field [0001] The invention relates to a method for recognizing temporal relationship of Chinese-Vietnamese news events based on a two-way cross-attention mechanism, and belongs to the technical field of natural language processing. Background technique [0002] Recognition of temporal relationship between events is a hot topic in current natural language processing research. Issues of common concern between China and Vietnam are increasing, and related news reports are also increasing. Identifying the temporal relationship between Chinese and Vietnamese bilingual news events can obtain the temporal logical relationship between bilingual news events, which will help relevant departments grasp the dynamics of the two countries and orientation, and make correct countermeasures. At present, the recognition of event temporal relationship requires a lot of manpower to design timeline-based templates. It is difficult to obtain the implicit semantic information in sentenc...

AI Technical Summary

Problems solved by technology

[0003] The present invention provides a Chinese-Vietnamese news event temporal relationship recognition method based on a two-way cross-attention mechanism, which is used to solve the current event temporal relationship recognition that requires a large amount of manpower to design a template based on the time axis, and the implicit semantic information in the sentence is difficult to obtain, and It is difficult to represent texts in different languages ​​in the same feature space, and it solves the problem of obtaining the time series relationship of cross-language news events.

Images