Entity relationship extraction method and device based on pre-training language model

Abstract

The invention provides an entity relationship extraction method and device based on a pre-training language model. The method comprises the steps of obtaining a corresponding candidate template from a to-be-extracted knowledge text library based on seed knowledge; the candidate templates are sorted and screened, and a Prompt template is determined; performing entity knowledge mining on original input information based on a prompt text generated by the Prompt template, a pre-training language model and a Prompt Tuning mode to obtain corresponding entity knowledge; and taking the entity knowledge as new seed knowledge, performing mining by utilizing the new seed knowledge to generate a new candidate template, and performing loop iteration processing based on the new candidate template to obtain a knowledge mining result output in the loop iteration processing process. According to the entity relation extraction method based on the pre-training language model, the labeling cost of the Prompt template can be reduced, and the knowledge mining efficiency is effectively improved.

Description

technical field [0001] The invention relates to the technical field of data analysis, in particular to a method and device for extracting entity relationships based on a pre-trained language model. In addition, it also relates to an electronic device and a processor-readable storage medium. Background technique [0002] In recent years, with the application of pre-trained language model (PLM) in various fields of natural language processing (NLP) more and more widely, natural language processing has also made great progress. However, as the number of parameters of the pre-training model increases, the hardware requirements and data labeling cost for fine-tuning it are getting higher and higher, and the problem of long cycle is gradually becoming more prominent. Therefore, those skilled in the art are in urgent need of a A more effective prompt-tuning scheme to improve the efficiency of knowledge mining based on pre-trained language models. However, the Prompt templates use...

AI Technical Summary

Problems solved by technology

[0003] To this end, the present invention provides a method and device for extracting entity relationships based on a pre-trained language model to solve the problem in the prior art that human participation is required for Prompt template labeling in knowledge-based mining schemes, resulting in poor knowledge mining efficiency.

Images