Text classification method and device based on heterogeneous neural network

Abstract

The invention discloses a text classification method and device based on a heterogeneous neural network, and the method comprises the steps: S1, constructing an N-layer text classification multi-way tree corresponding to a tree structure in combination with the tree structure of an actual classification system in a training data set, and according to the structure of the N-layer text classification multi-way tree, respectively writing training data in the training data set into classification files corresponding to each level, performing word segmentation on Chinese texts of each classification file, performing feature selection, and storing the selected features into corresponding feature files; S2, constructing a text heterogeneous neural network corresponding to the tree structure; and S3, setting heterogeneous neural network algorithm parameters, carrying out parameter adjustment, iteratively generating and storing each classification and sub-classification model of training data, using a verification data set for carrying out accuracy judgment. The categories and the hierarchical relation between the categories are also added into the heterogeneous neural network, and the problems of vector representation and link data sparsity are learned in a category display manner.

Description

technical field [0001] The present invention relates to the fields of natural language processing and artificial intelligence, in particular to a text classification method and device based on a heterogeneous neural network. Background technique [0002] Text classification is a process in which computers automatically classify natural language according to a certain classification system, and it has applications in many fields. With the increasing number of application scenarios, scenarios with a large classification system are becoming more and more common. Taking the customer service center platform as an example, merchants hope to achieve accurate classification and quick recording of customer visit work orders through automatic text classification, so as to speed up on-site The energy efficiency of operator service response improves the overall service quality; it optimizes management by improving the intelligence of the overall operation data analysis of the hotline. ...

AI Technical Summary

Problems solved by technology

However, due to the complex business logic, the classification system in this scenario is very large, which is different from traditional text classification problems

Images