Intention identification method and device, computer equipment, and computer-readable storage medium

Abstract

The invention relates to an intention identification method and device, computer equipment, and a computer storage medium. The method is characterized in that by introducing a three-layer automatic judgment mechanism, intention identification matching on text information of a to-be-recognized intention is carried out, so as to have intention identification matching conditions from strict to tolerant. The intention identification accuracy is ensured through complete regular matching; under the condition that a complete matching layer cannot match the text information of the to-be-recognized intention, the text of the to-be-recognized intention is input into an intention identification model layer which is completely trained for matching, and the generalization performance of the intention identification is fully ensured; finally, under the condition that the intention identification of the intention identification model layer fails, fuzzy identification is carried out by using the regular matching rule again, so that the intention identification is ensured to be accurately and comprehensively realized, and the problem that the intention identification cannot give consideration to generalization performance and accuracy in the prior art is solved.

Description

technical field [0001] The present application relates to the field of machine identification, in particular to an intent identification method, device, computer equipment and computer-readable storage medium. Background technique [0002] At present, the methods of user intent recognition are mainly divided into two categories. The first category is the method based on regular matching. Although using this method for intent recognition is more accurate, the generalization ability is poor, which usually results in relaxing some matching rules. Matching the wrong intentions leads to a large number of intention recognition must be based on artificially writing a large number of matching rules, resulting in huge labor costs; the other is based on deep learning, the generalization performance of deep learning algorithms is greatly improved, but usually often Identify false intent on some simple sentences. [0003] In related technologies, user intent recognition cannot take int...

AI Technical Summary

Problems solved by technology

[0002] At present, the methods of user intent recognition are mainly divided into two categories. The first category is the method based on regular matching. Although using this method for intent recognition is more accurate, the generalization ability is poor, which usually results in relaxing some matching rules. Matching the wrong intentions leads to a large number of intention recognition must be based on artificially writing a large number of matching rules, resulting in huge labor costs; the other is based on deep learning, the generalization performance of deep learning algorithms is greatly improved, but usually often Identify false intent on some simple sentences

[0003] In related technologies, user intent recognition cannot take into account both generalization performance and accuracy, and no effective solution has been proposed yet

Images