Weakly-Supervised Neural Text Classification

Abstract

Deep neural networks are gaining increasing popularity for the classic text classification task, due to their strong expressive power and less requirement for feature engineering. Despite such attractiveness, neural text classification models suffer from the lack of training data in many real-world applications. Although many semi-supervised and weakly-supervised text classification models exist, they cannot be easily applied to deep neural models and meanwhile support limited supervision types. In this paper, we propose a weakly-supervised method that addresses the lack of training data in neural text classification. Our method consists of two modules: (1) a pseudo-document generator that leverages seed information to generate pseudo-labeled documents for model pre-training, and (2) a self-training module that bootstraps on real unlabeled data for model refinement. Our method has the flexibility to handle different types of weak supervision and can be easily integrated into existing deep neural models for text classification. We have performed extensive experiments on three real-world datasets from different domains. The results demonstrate that our proposed method achieves inspiring performance without requiring excessive training data and outperforms baseline methods significantly.

arXiv commentCIKM 2018 Full Paper

References in corpus (7)

• Neural Machine Translation by Jointly Learning to Align and Translate
• Character-level Convolutional Networks for Text Classification
• PTE: Predictive Text Embedding through Large-scale Heterogeneous Text Networks
• Text Understanding from Scratch
• Adversarial Training Methods for Semi-Supervised Text Classification
• Effective Use of Word Order for Text Categorization with Convolutional Neural Networks
• Directional Statistics in Machine Learning: a Brief Review

Cited by in corpus (37)

• Deep Learning Based Text Classification: A Comprehensive Review
• BOND: BERT-Assisted Open-Domain Named Entity Recognition with Distant Supervision
• Hierarchical Text Classification with Reinforced Label Assignment
• Discriminative Topic Mining via Category-Name Guided Text Embedding
• Hierarchical Topic Mining via Joint Spherical Tree and Text Embedding
• Topic Discovery via Latent Space Clustering of Pretrained Language Model Representations
• Denoising Multi-Source Weak Supervision for Neural Text Classification
• CoAnnotating: Uncertainty-Guided Work Allocation between Human and Large Language Models for Data Annotation
• MixText: Linguistically-Informed Interpolation of Hidden Space for Semi-Supervised Text Classification
• TaxoCom: Topic Taxonomy Completion with Hierarchical Discovery of Novel Topic Clusters
• Efficient Path Prediction for Semi-Supervised and Weakly Supervised Hierarchical Text Classification
• Weakly-Supervised Scientific Document Classification via Retrieval-Augmented Multi-Stage Training
• A Survey on Retrieval And Structuring Augmented Generation with Large Language Models
• Semi-Supervised Models via Data Augmentationfor Classifying Interactive Affective Responses
• vONTSS: vMF based semi-supervised neural topic modeling with optimal transport
• Text Classification Using Label Names Only: A Language Model Self-Training Approach
• The Rich Get Richer: Disparate Impact of Semi-Supervised Learning
• X-Class: Text Classification with Extremely Weak Supervision
• Fine-Tuning Pre-trained Language Model with Weak Supervision: A Contrastive-Regularized Self-Training Approach
• A Variational Approach to Weakly Supervised Document-Level Multi-Aspect Sentiment Classification
• WOT-Class: Weakly Supervised Open-world Text Classification
• Named Entity Recognition with Small Strongly Labeled and Large Weakly Labeled Data
• Minimally Supervised Categorization of Text with Metadata
• A Variational Approach to Unsupervised Sentiment Analysis
• ViF-SD2E: A Robust Weakly-Supervised Method for Neural Decoding
• Seed Word Selection for Weakly-Supervised Text Classification with Unsupervised Error Estimation
• Coarse2Fine: Fine-grained Text Classification on Coarsely-grained Annotated Data
• Out-of-Scope Intent Detection with Self-Supervision and Discriminative Training
• Weakly-supervised Text Classification Based on Keyword Graph
• Minimally-Supervised Structure-Rich Text Categorization via Learning on Text-Rich Networks
• Customer Sentiment Analysis using Weak Supervision for Customer-Agent Chat
• Weakly-Supervised Dense Action Anticipation
• User-Guided Aspect Classification for Domain-Specific Texts
• Distantly-Supervised Named Entity Recognition with Noise-Robust Learning and Language Model Augmented Self-Training
• TEAMMix: Taxonomy Enrichment Augmentation and Minority-augmented Mixing Strategy for LLM-enhanced Weak-Supervised Hierarchical Text Classification
• Out-of-Category Document Identification Using Target-Category Names as Weak Supervision
• Variational Weakly Supervised Sentiment Analysis with Posterior Regularization