Intention recognition method and device, electronic equipment and storage medium

Abstract

The invention provides an intention recognition method and device, an electronic device and a storage medium. The intention recognition method comprises the following steps: obtaining to-be-recognizedinformation; identifying the to-be-identified information to obtain intention vectors corresponding to the to-be-identified information, wherein the intention vector comprises a plurality of characterization values, and each characterization value at least corresponds to one preset intention; judging an intention corresponding to the to-be-identified information according to the plurality of characterization values; and if the plurality of characterization values are all smaller than or equal to the judgment value, determining that the to-be-identified information is unintentional information. The invention discloses an intention recognition method and device, electronic equipment and a storage medium, and in application, the method comprises carrying out the recognition of to-be-recognized information, obtaining an intention vector, each representation value in the intention vector corresponding to a preset intention, and all preset intentions being forward intentions; when performing recognition of the to-be-recognized information, analyzing and calculating the representational value of the preset intention only; therefore, the complexity of the operation is reduced, the operation efficiency is improved, and the accuracy is also improved.

Description

technical field [0001] The present invention relates to the field of natural language understanding, in particular to an intention recognition method, device, electronic equipment and storage medium. Background technique [0002] The dialogue system is a commonly used functional system. Unintentional negative samples are added to the training of the system for training. The trained system can simultaneously improve the recognition rate of positive and negative intentions. [0003] In the common method, in the multi-intent category recognition problem, in addition to the N (N>1) intent categories that business cares about, it is often accompanied by unlabeled and unintentional text input. In the past, the unintentional category was regarded as "other " category is added to the label system and becomes an N+1 classification task. When predicting, it is judged that if it is "other", it is an unintentional input. However, when calculating the intent vector, it is also necess...

AI Technical Summary

Problems solved by technology

But in this way, when calculating the intent vector, it is also necessary to calculate the value corresponding to the "other" class, which will increase the complexity of the intent vector calculation and affect the calculation efficiency. At the same time, it will also increase the establishment of the label system high difficulty

Images