Combined neural network-based text classification method, device and computer equipment

Abstract

The invention relates to a combined neural network-based text classification method, a combined neural network-based text classification device, computer equipment and a storage medium. The method comprises the steps of inputting to-be-classified texts into a trained first text classification model, and determining a first classification confidence coefficient, wherein the first text classification model is a neural network obtained by training based on a refined data set; if the first classification confidence coefficient is smaller than a first confidence coefficient threshold value, inputting the to-be-classified text into a trained second text classification model, and determining a second classification confidence coefficient; wherein the second text classification model is a neural network obtained by training based on a full data set; and if the second classification confidence is greater than or equal to the second confidence threshold, determining the classification result ofthe second text classification model as the classification result of the to-be-classified text. By adopting the combined neural network to perform intention classification on the to-be-classified text, the high accuracy of a classification result of the to-be-classified text is ensured. The reliability of the intention classification result is good. The automatic classification of text intentionscan be realized, and the application range is wide.

Description

technical field [0001] The present application relates to the technical field of artificial intelligence, in particular to a text classification method, device, computer equipment and storage medium based on a combined neural network. Background technique [0002] With the development of Internet technology, the amount of text data has increased rapidly. Text classification is an important content of text mining. It refers to determining the category of each text in the text collection to be processed according to the predefined categories. This technology can be applied in various intelligent interaction scenarios. For example, in the intelligent collection business, the intelligent robot needs to make a corresponding response according to the message replied by the customer. If the message replied by the customer is different, the intelligent robot needs to reply with different speech skills. Specifically, for example, the robot asks the customer "Your loan of 2,000 yuan ...

AI Technical Summary

Problems solved by technology

However, due to the diversity and complexity of text corresponding intents, when classifying text intents based on existing text classification models, the reliability of the obtained intent classification results is not ideal, and it is difficult to automatically classify text intents

Images