Unlock instant, AI-driven research and patent intelligence for your innovation.

Mongolian-Chinese machine translation method based on tree-to-sequence

A machine translation and sequence technology, applied in the field of machine translation to achieve the effect of improving translation accuracy

Inactive Publication Date: 2020-01-24
INNER MONGOLIA UNIV OF TECH
View PDF1 Cites 3 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, existing NMT models do not allow to perform this alignment

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Mongolian-Chinese machine translation method based on tree-to-sequence
  • Mongolian-Chinese machine translation method based on tree-to-sequence
  • Mongolian-Chinese machine translation method based on tree-to-sequence

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0033] The implementation of the present invention will be described in detail below in conjunction with the drawings and examples.

[0034] figure 1 A pair of parallel sentences in Chinese and Mongolian is shown. Chinese and Mongolian are linguistically distant in many respects, they have different syntactic structures, and words and phrases are defined in different lexical units. The present invention aims at incorporating syntactic components of the known source language into the model using a light alignment algorithm to improve word alignment and translation accuracy.

[0035] In order to achieve the above purpose, the present invention uses an attentional NMT model to utilize syntactic information, and still uses the encoder-decoder model as the overall framework of the translation process, after the phrase structure of the source sentence, recursively in a bottom-up manner Encode a sentence to produce a vector representation of the sentence and decode it while aligni...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a Mongolian-Chinese machine translation method based on a tree-to-sequence. A tree-to-sequence NMT model is adopted; a sequence with a source side phrase structure is expandedto a sequence model; a self-attention mechanism is added into the model, the self-attention mechanism not only enables a decoder to actively inquire the most relevant information in each step, but also greatly shortens the information flow distance, and in addition, the decoder can be aligned with phrases of source statements and words while generating translated words. Experimental results of a 1200000 Mongolian Chinese bilingual parallel corpus data set show that the model of the invention is obviously superior to a sequence-to-sequence attention NMT model, and is better than the most advanced tree-to-string SMT system.

Description

technical field [0001] The invention belongs to the technical field of machine translation, in particular to a tree-to-sequence-based Mongolian-Chinese machine translation method. Background technique [0002] Machine translation (MT) has been one of the most complex language processing problems, and recent advances in neural machine translation (NMT) have made translation possible using simple end-to-end architectures. [0003] In the encoder-decoder model, the encoder reads the entire sequence of source words to produce a fixed-length vector, and the decoder generates target words from the vector. Encoder-decoder models have been extended with attention mechanisms, which allow the models to jointly learn a soft alignment between source and target languages. NMT models achieve state-of-the-art results on English-French and English-German translation tasks. However, it remains to be seen whether NMT is competitive with traditional statistical machine translation (SMT) meth...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06F40/58G06F40/289G06F40/12G06N3/04
CPCG06N3/044
Inventor 苏依拉薛媛赵旭卞乐乐范婷婷张振
Owner INNER MONGOLIA UNIV OF TECH