Multi-label Text Classification using GloVe and Neural Network Models
Abstract
This study addresses the challenges of multi-label text classification. The difficulties arise from imbalanced data sets, varied text lengths, and numerous subjective feature labels. Existing solutions include traditional machine learning and deep neural networks for predictions. However, both approaches have their limitations. Traditional machine learning often overlooks the associations between words, while deep neural networks, despite their better classification performance, come with increased training complexity and time. This paper proposes a method utilizing the bag-of-words model approach based on the GloVe model and the CNN-BiLSTM network. The principle is to use the word vector matrix trained by the GloVe model as the input for the text embedding layer. Given that the GloVe model requires no further training, the neural network model can be trained more efficiently. The method achieves an accuracy rate of 87.26% on the test set and an F1 score of 0.8737, showcasing promising results.
Cite
@article{arxiv.2312.03707,
title = {Multi-label Text Classification using GloVe and Neural Network Models},
author = {Hongren Wang},
journal= {arXiv preprint arXiv:2312.03707},
year = {2024}
}
Comments
I have discovered some significant modifications that need to be made. These modifications are crucial and will substantially impact the results of the study. I plan to thoroughly revise the paper and resubmit it to ensure the accuracy and completeness of the research findings