Single-label text classification method, computing device and computer readable storage medium

Abstract

The invention provides a single-label text classification method, computing equipment and a computer readable storage medium. The method comprises the steps: constructing a sample set based on user behavior data and commodity data, wherein the sample set comprises a plurality of samples, and each sample comprises a first field indicating text and a second field indicating a single label; performing word segmentation on the first field of each sample to obtain at least one feature word, and obtaining a formatted sample of the sample based on the second field and the at least one feature word; training a first neural network model based on a plurality of formatted samples of the plurality of samples; dividing the plurality of formatted samples into a plurality of groups based on the second field; and for each group, determining a second neural network model based on the average word vector of at least one feature word in the group. A large-data-volume training sample set is constructed by utilizing data from multiple sources, and the performance loss of an online system is reduced by utilizing a combined model.

Description

technical field [0001] The present disclosure generally relates to the field of machine learning, and more specifically, relates to a single-label text classification method, a computing device, and a computer-readable storage medium. Background technique [0002] Currently, various machine learning-based methods have been proposed to classify objects such as text and pictures. These classification methods usually require large amounts of data to train neural network models to classify various objects into multiple categories. However, in some cases, such as e-commerce search scenarios, the text to be classified (search term) is usually a single-label text, which often faces problems such as insufficient data volume and excessive data noise. Especially in scenarios such as system cold start, user behavior data is insufficient. Under such conditions, how to construct an effective data set is the cornerstone of producing an effective model and achieving accurate text classif...

AI Technical Summary

Problems solved by technology

[0003] In addition, conventional classification methods usually use a deep network model. Due to the deep network layer, complex structure, and various parameters, the deployment of the model online will cause a certain degree of performance loss. This part of the loss is sensitive to response time systems (such as search , recommendation system, etc.) can not be ignored

[0004] Furthermore, in the case of a small data volume, the single-model classification scheme makes the entire classification algorithm completely dependent on the quality of a single model, which lacks robustness

Images