Customer reviews play a crucial role in assessing customer satisfaction,
gathering feedback, and driving improvements for businesses. Analyzing these
reviews provides valuable insights into customer sentiments, including
compliments, comments, and suggestions. Text classification techniques enable
businesses to categorize customer reviews into distinct categories,
facilitating a better understanding of customer feedback. However, challenges
such as overfitting and bias limit the effectiveness of a single classifier in
ensuring optimal prediction. This study proposes a novel approach to address
these challenges by introducing a stacking ensemble-based multi-text
classification method that leverages transformer models. By combining multiple
single transformers, including BERT, ELECTRA, and DistilBERT, as base-level
classifiers, and a meta-level classifier based on RoBERTa, an optimal
predictive model is generated. The proposed stacking ensemble-based multi-text
classification method aims to enhance the accuracy and robustness of customer
review analysis. Experimental evaluations conducted on a real-world customer
review dataset demonstrate the effectiveness and superiority of the proposed
approach over traditional single classifier models. The stacking ensemble-based
multi-text classification method using transformers proves to be a promising
solution for businesses seeking to extract valuable insights from customer
reviews and make data-driven decisions to enhance customer satisfaction and
drive continuous improvement.

通过引入基于堆叠集成的多文本分类方法来解决单一分类器的欠拟合和偏差限制，结合 BERT、ELECTRA、DistilBERT、RoBERTa 等多个转换器模型，在真实客户评论数据集上进行实验评估，验证了该方法在提高准确性和鲁棒性方面的有效性和优越性，证明了使用转换器的堆叠集成的多文本分类方法是从客户评论中提取有价值的见解并作出数据驱动决策以提升客户满意度和推动持续改进的有前途的解决方案。

优化多类文本分类：基于变形金刚的多样性叠加集成框架

Optimizing Multi-Class Text Classification: A Diverse Stacking Ensemble  Framework Utilizing Transformers

Automatic tagging of knowledge points for practice problems is the basis for
managing question bases and improving the automation and intelligence of
education. Therefore, it is of great practical significance to study the
automatic tagging technology for practice problems. However, there are few
studies on the automatic tagging of knowledge points for math problems. Math
texts have more complex structures and semantics compared with general texts
because they contain unique elements such as symbols and formulas. Therefore,
it is difficult to meet the accuracy requirement of knowledge point prediction
by directly applying the text classification techniques in general domains. In
this paper, K12 math problems taken as the research object, the LABS model
based on label-semantic attention and multi-label smoothing combining textual
features is proposed to improve the automatic tagging of knowledge points for
math problems. The model combines the text classification techniques in general
domains and the unique features of math texts. The results show that the models
using label-semantic attention or multi-label smoothing perform better on
precision, recall, and F1-score metrics than the traditional BiLSTM model,
while the LABS model using both performs best. It can be seen that label
information can guide the neural networks to extract meaningful information
from the problem text, which improves the text classification performance of
the model. Moreover, multi-label smoothing combining textual features can fully
explore the relationship between text and labels, improve the model's
prediction ability for new data and improve the model's classification
accuracy.

本研究提出 LABS 模型，结合标签 - 语义注意力和多标签平滑技术，旨在改进对数学问题知识点的自动标记。实验证明，LABS 模型应用于 K12 数学问题时，利用文本分类技术和数学文本的特殊元素能够最大化地提高模型的分类准确性及预测性能。