Concept tree-based intention identification method and apparatus, and computer device

Abstract

The invention provides a concept tree-based intention identification method and apparatus, and computer device, relating to the technical field of data analysis. By performing concept tree construction, the concept tree comprises at least one layer; each layer comprises at least one intention; wherein each intention corresponds to at least one keyword; wherein each keyword is provided with a corresponding weight; carrying out word segmentation processing on the target statement to obtain at least one traversal word, aiming at each current traversal word; traversing the keyword corresponding toeach intention in the concept tree; calculating the word vector similarity between the current traversal word and each traversed keyword, calculating the intention score corresponding to each intention of the target statement according to the word vector similarity and the weight of each keyword, and determining the intention of the target statement according to the intention score and the intention threshold of each layer. According to the method, a large number of training samples in a specific field are not needed, relation network analysis is achieved through the concept tree, and therefore the intention of the target statement can be accurately estimated.

Description

technical field [0001] The present invention relates to the technical field of data analysis, in particular to an intention recognition method, device and computer equipment based on a concept tree. Background technique [0002] In the field of service robots, robots can handle business consulting services in specific fields. In the process of processing consulting services, whether the robot can correctly identify the user's intention is an important factor for the robot to effectively provide users with business questions and answers. Therefore, the research based on intent recognition is constantly developing. [0003] At present, in dealing with intent recognition in a specific field, most of them use machine learning or deep learning methods to train models on data to classify intents. However, because the application scenario is intent recognition in a specific field, there are relatively few training samples available, so the method based on machine learning or deep ...

AI Technical Summary

Problems solved by technology

However, because the application scenario is intent recognition in a specific field, there are relatively few training samples available, so the method based on machine learning or deep learning may be due to the lack of data training samples in a specific field. The effect of the trained model is not obvious or the model is not available at all

Images