Weakly-supervised text classification

Meng, Yu

Permalink

https://hdl.handle.net/2142/104867 Copy

Description

Title

Weakly-supervised text classification

Author(s)

Meng, Yu

Issue Date

2019-04-22

Director of Research (if dissertation) or Advisor (if thesis)

Han, Jiawei

Department of Study

Computer Science

Discipline

Computer Science

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• Text Classification
• Weakly-supervised Learning
• Neural Classification Model
• Hierarchical 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 work, we propose a weakly-supervised framework that addresses the lack of training data in neural text classification. Our framework 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 framework has the flexibility to handle different types of weak supervision and can be easily integrated into existing deep neural models for text classification. Based on this framework, we propose two methods, WeSTClass and WeSHClass, for flat text classification and hierarchical text classification, respectively. We have performed extensive experiments on real-world datasets from different domains. The results demonstrate that our proposed framework achieves inspiring performance without requiring excessive training data and outperforms baselines significantly.

Graduation Semester

2019-05

Type of Resource

text

Permalink

http://hdl.handle.net/2142/104867

Copyright and License Information

Copyright 2019 Yu Meng

Owning Collections