Fine-grained named entity recognition method based on pre-training coding feature hierarchical processing

Abstract

The invention discloses a fine-grained named entity recognition method based on pre-training coding feature hierarchical processing, and relates to the technical field of natural language processing, and the method comprises the steps: obtaining a Chinese pre-training coding vector through employing a pre-training model ALBERT, then carrying out the hierarchical extraction of useful bidirectional context features in the pre-training features, and fusing the features. And then the objective function is accurately optimized by utilizing the advantages of optimal solution solving of a supervised learning technology and a conditional random field CRF, so that model parameters are automatically learned to improve the recognition effect of the model on a to-be-recognized data set. Experiments show that by adopting the method of the invention, the problem of fine-grained named entity recognition can be effectively solved, and a better recognition effect is achieved.

Description

technical field [0001] The present invention relates to the technical field of natural language processing, specifically to the technical field of named entity recognition, and more specifically to a fine-grained named entity recognition method based on layered processing of pre-trained coding features. Background technique [0002] BERT is a large-scale pre-trained language model based on Transformers encoder. In recent years, BERT has demonstrated strong strength in many downstream tasks. Named Entity Recognition (NER) is a basic task in natural language processing with a wide range of applications. [0003] The named entity recognition method of traditional machine learning, which formalizes the named entity recognition task into a sequence labeling task, predicts the label of each word or word, and jointly predicts the entity boundary and entity type, which is heavily dependent on the dictionary library and cannot be recognized. There are shortcomings such as entity ne...

AI Technical Summary

Problems solved by technology

[0003] The named entity recognition method of traditional machine learning, which formalizes the named entity recognition task into a sequence labeling task, predicts the label of each word or word, and jointly predicts the entity boundary and entity type, which is heavily dependent on the dictionary library and cannot be recognized. Disadvantages such as entity nesting, inability to characterize words or polysemy of words, and lack of migration

Images