Method and device for remote supervision relation extraction based on consistent text enhancement

Abstract

The present invention provides a method and device for remote supervision relation extraction based on consistent text enhancement, the method comprising: dividing multiple sentence instances according to entity pairs and relation labels to obtain multiple sentence packages; Each sentence instance uses a different text enhancement method to obtain the strong enhancement samples and weak enhancement samples corresponding to each sentence instance in each sentence bag; determine the noise sample, and use the strong enhancement of the irrelevant sentence instance and the noise sample The samples and weakly enhanced samples train the relationship prediction model to obtain a trained relationship prediction model; use the trained relationship prediction model to predict the sentence package to be predicted, and obtain the corresponding relationship label. Through the present invention, the size of the data set can be increased through consistent text enhancement, the generalization learning ability of the model can be enhanced, and more supervision information can be learned by the "NA" category and the noise sample constraint model.

Description

technical field [0001] The present invention relates to the field of natural language processing, in particular to a method and device for extracting long-distance supervision relations based on consistent text enhancement. Background technique [0002] From the massive information on the Internet, a large amount of valuable knowledge and information can be extracted through the relevant technology of information extraction. As an important link in information extraction, Relation Extraction (RE) aims to extract the relationship between entities from text, for other natural language applications such as building knowledge graphs, search engines, dialogue generation, natural question answering, information retrieval, etc. Provided significant support. [0003] The training of relation extraction models requires a large number of labeled samples to provide supervision information. However, the same relation type may have different textual expressions, and at the same time, d...

AI Technical Summary

Problems solved by technology

[0008] 2) Many research methods reduce the weight of noise samples in the training set or directly filter them out, so that the effective information contained in these noise samples cannot be fully utilized;

[0009] 3) The disturbance added by methods such as confrontation generation, although it can increase the anti-disturbance ability of the model, it usually cannot provide disturbances that meet the actual situation, is not stable, and tends to deviate the direction of model training

Images