BERT fusion capsule network elevator fault complaint text classification method

Abstract

The invention discloses an elevator fault complaint text classification method based on BERT fused with a capsule network, and the method comprises the steps: firstly carrying out the semantic feature extraction of a text through the powerful word and sentence expression capability of a pre-training language model BERT and the capability of extracting a long-distance dependency relationship, and obtaining a vector corresponding to a first character in the last output layer of the BERT; the complaint text sentence vector is processed by a linear layer and a Tanh activation layer to obtain complaint text sentence vector representation, so that the sentence vector dimension is reduced, and the classification efficiency is improved; then, a mixed sampling method combining oversampling and undersampling is used for solving the problem of unbalanced data category distribution; and finally, secondary extraction and fusion of features are realized by means of a capsule network, local and overall relationships of the texts are obtained, word sequences and semantics are reserved, intelligent classification of elevator fault complaint texts is finally realized, reference values are provided for regular maintenance of elevators, and a solution is also provided for efficient utilization of fault complaint text data.

Description

technical field [0001] The invention relates to a text classification method for elevator fault complaints based on BERT fusion capsule network. Background technique [0002] In 2021, the number of elevators in China will reach 8.447 million units, a year-on-year increase of 5.6%, and the safe operation and maintenance of elevators has become a crucial issue. The smart elevator management system is an online elevator monitoring and maintenance service system, which can realize the functions of elevator real-time monitoring, fault event recording, elevator maintenance management and user complaints. The problem of handling user complaints is to ensure the safe operation of elevators. dimensional key. User complaint information often exposes problems or faults in the operation of elevators, and plays an important guiding role in adjusting the decision-making of elevator-related units and improving elevator quality and service. Traditionally, elevator user complaint texts are...

AI Technical Summary

Problems solved by technology

However, when the capsule network is used for text classification, there is a problem of lack of long-distance dependencies in the encoded text.

Images