Supercharge Your Innovation With Domain-Expert AI Agents!

Named entity recognition method based on cascading model

A technology for named entity recognition and named entities, which is applied in neural learning methods, biological neural network models, special data processing applications, etc., can solve complex Chinese named entity recognition and other problems, and achieve reliable classification, improved success rate, and accurate grasp Effect

Active Publication Date: 2019-08-09
NANJING UNIV
View PDF4 Cites 15 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

This method solves the recognition problem of complex Chinese named entities in the Internet text environment

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
  • Named entity recognition method based on cascading model
  • Named entity recognition method based on cascading model
  • Named entity recognition method based on cascading model

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0016] as attached figure 1 As shown, the BiLSTM-CRF named entity model structure consists of a distributed embedding layer, a deep neural network layer and a conditional random field layer. The distributed embedding module uses word2vec to train word vectors. This method associates the distributed representation of text with the meaning between words, eliminating the phenomenon of word gaps. Using pre-trained word vectors as the input of deep learning to deal with natural language problems has become a classic and mature method. Many works have shown that using pre-trained word vectors compared with random embeddings, the entire neural network converges faster; the trained model has a greater improvement in accuracy and recall; especially in small data The advantages of using word2vec are more obvious in this case.

[0017] For information sequences, the information has complex temporal correlation with each other, and more importantly, for named entity recognition tasks, ...

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

A complex Chinese named entity recognition method based on a cascading model comprises the following steps: 1) a model training stage: a, training and storing a low-level BiLSTM-CRF model under the calculation of an improved loss function through a marked named entity corpus set; b, training and storing a high-level BiLSTM-CRF model through labeled named entity recognition corpus; and 2) a model prediction stage: sending the corpus with prediction into a low-level model, identifying a coarse-grained named entity as a preliminary result, and sending the preliminary result into a high-level model. The high-level model continues to recognize the preliminary result, if the recognition result is not a single named entity, the result is input into the high-level model again, and all the resultsare known as single named entities; and 3) outputting a result: collecting all named entities, namely all named entities output by the high-level network, obtained by processing the corpus through thecascading model, and taking the named entities as a final result identified in the whole identification process.

Description

technical field [0001] The invention relates to a named entity recognition method based on a layered model, and the method solves the recognition problem of complicated Chinese named entities under the Internet text environment. Background technique [0002] Natural Language Processing (NLP) technology is a subfield of computer information engineering. The goal is to process and analyze massive text data, so that computer programs can use lexical, grammatical, semantic and other information to complete recognition, understanding and output of natural language texts. Tasks such as word segmentation, named entity recognition, relation extraction, machine translation, natural language generation, question answering systems, sentiment analysis, and more. Natural language processing technology is becoming more and more mature under the exploration and research of methods such as rule learning and statistical learning. In recent years, representation learning and deep neural netw...

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): G06F17/27G06N3/04G06N3/08
CPCG06N3/08G06F40/295G06N3/045
Inventor 吴骏顾溢张哲成谈志文李宁
Owner NANJING UNIV
Features
  • R&D
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2025 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com