Multi-feature fusion Vietnamese keyword generation method

Abstract

The invention relates to a multi-feature fusion Vietnamese keyword generation method, and belongs to the field of natural language processing. The Vietnamese keyword generation comprises the following steps: carrying out keyword prediction on a Vietnamese news text to obtain a keyword of highly summarizing news text information; according to the method, part-of-speech information, named entity information and position information are fused in the encoding process; secondly, the guiding effect of the title information in the generation process is enhanced by using a bidirectional attention mechanism; and finally, the feature vector fused with various semantic information is sent into a decoding layer, final prediction probability distribution is output, and further Vietnamese keywords are generated. According to the method, a good effect is achieved in Vietnamese keyword generation, and support is provided for subsequent text classification and information retrieval.

Description

technical field [0001] The invention relates to a multi-feature fusion Vietnamese keyword generation method, which belongs to the field of natural language processing. Background technique [0002] In natural language processing tasks, keywords are usually used as multi-word units to summarize the basic ideas of documents in short texts. Vietnamese keyword generation provides downstream tasks such as Chinese-Vietnamese bilingual text summarization, cross-language information retrieval, and public opinion analysis in Southeast Asia. important support. [0003] Good progress has been made in keyword generation in the English environment. Meng Rui et al. proposed a CopyRNN network using an encoder-decoder structure, an attention mechanism and a replication mechanism, and trained a generative model on a large-scale corpus. Bidirectional RNNs with gated recurrent units are not as good as non-deep learning methods at extracting keywords that already appear in documents in most da...

AI Technical Summary

Problems solved by technology

[0006] The present invention provides a Vietnamese keyword generation method with multi-feature fusion, which is used to generate keywords for Vietnamese documents, and solves the problem of insufficient keyword generation due to the scarcity of Vietnamese and other low-resource language corpora and insufficient features extracted by the model. exact question

Images