Text representation method and device and computer equipment

Abstract

The embodiment of the invention discloses a text representation method and device and computer equipment. The text characterization method comprises the steps of performing word segmentation on an input sentence text to obtain a first word vector matrix and a first character vector matrix; dynamically combining the first word vector matrix and the first character vector matrix through an attentionmechanism to obtain a word information background matrix and a character information background matrix; performing matrix splicing on the first word vector matrix and the word information backgroundmatrix to obtain a second word vector matrix, and performing matrix splicing on the first word vector matrix and the word information background matrix to obtain a second word vector matrix; and thencarrying out deep encoding on the second word vector matrix and the second character vector matrix to obtain a word matrix layer and a character matrix layer, and carrying out matrix splicing to obtain a complete encoding layer of the sentence text as a representation result corresponding to the sentence text. According to the text characterization method, sentence semantic information can be expressed more completely and accurately.

Description

technical field [0001] The invention relates to the field of natural language processing, in particular to a text representation method, device and computer equipment. Background technique [0002] Using natural language to communicate with computers has been pursued by people for a long time. With the rapid development of deep learning technology, more and more deep learning technologies have been applied to various fields of natural language processing, and a variety of natural language models based on deep learning have also been proposed. How to better represent natural language models has become very important issue. [0003] Some models analyze a word obtained by a word segmentation tool as a semantic unit of a sentence when modeling a sentence in a text. For Chinese, a word is directly used as a semantic representation, ignoring the information between characters within a word. Moreover, the Chinese word segmentation results of the word segmentation tool are not co...

AI Technical Summary

Problems solved by technology

Moreover, the Chinese word segmentation results of the word segmentation tool are not completely accurate, so analyzing a word as a semantic unit of a sentence cannot fully express the meaning of the sentence

Other models represent individual characters as a semantic unit, but a single character in Chinese is ambiguous and cannot accurately represent the current contextual information

There are also some models that directly combine Chinese characters and word information to represent, but for a word, some characters can express the semantics of the entire word, while some characters will not affect the semantics of the word, and cannot accurately represent current context information

[0004] It can be seen that the existing natural language model cannot reflect the importance of word-to-word and word-to-word for the combination of words and characters, and cannot accurately and perfectly represent the semantics of sentences

Images