Low-resource text recognition algorithm based on semantic elements

Abstract

The invention provides a low-resource text recognition algorithm based on semantic elements, and belongs to the technical field of natural language understanding. The method comprises the steps of obtaining a text sentence, and performing encoding processing on the text sentence to obtain encoded sentence tensor representation; performing semantic element recognition processing on the sentence tensor representation to obtain a semantic element recognition result; scaling the sentence tensor representation by using a semantic element recognition result; processing the scaled sentence tensor representation by using a mean pooling method to obtain a semantic element vector representation; processing the sentence tensor representation by using the mean pooling method to obtain sentence vectorrepresentation; splicing the sentence vector representation and the semantic element vector representation to obtain final sentence representation; and processing the final representation of the sentence to obtain a final text type probability. According to the invention, a semantic element recognition task is introduced so that the model has the capability of recognizing different semantic elements, and learning difficulty of the instruction text classification task is greatly reduced.

Description

technical field [0001] The invention belongs to the technical field of natural language understanding, and in particular relates to a low-resource text recognition algorithm based on semantic elements. Background technique [0002] In recent years, deep learning models have achieved remarkable results in many natural language understanding tasks. However, deep learning-based methods often require a relatively large amount of labeled data. Moreover, the application of natural language understanding has relatively strong scene-based characteristics, and it is impossible to directly develop applications using public corpus resources. In many fields and different scenarios, the cost of corpus labeling is relatively high. [0003] Voice command control is a human-computer interaction method that uses a voice control system. The usual implementation method is to use voice recognition technology to convert voice information into text, and then use text classification technology to ...

AI Technical Summary

Problems solved by technology

In this scenario, there are relatively few language expressions, but the accuracy of the classification model is high.

At present, those skilled in the art mostly use neural network language models such as BERT for processing. This model can ensure the convergence and generalization of the model with less data, but it is difficult to meet the accuracy requirements of the model, and it will also make the model too large to be able to Deploy offline on mobile

Images