Semi-supervised Text Regression with Conditional Generative Adversarial Networks

TL;DR

This paper introduces a semi-supervised text regression model using conditional GANs that effectively handles unbalanced datasets and predicts social outcomes directly from textual data.

Contribution

It presents a novel end-to-end semi-supervised text regression framework based on conditional GANs, suitable for limited labeled data scenarios.

Findings

Effective on unbalanced datasets with limited labels

End-to-end prediction without high-level feature selection

Potential for social and economic text analysis

Abstract

Enormous online textual information provides intriguing opportunities for understandings of social and economic semantics. In this paper, we propose a novel text regression model based on a conditional generative adversarial network (GAN), with an attempt to associate textual data and social outcomes in a semi-supervised manner. Besides promising potential of predicting capabilities, our superiorities are twofold: (i) the model works with unbalanced datasets of limited labelled data, which align with real-world scenarios; and (ii) predictions are obtained by an end-to-end framework, without explicitly selecting high-level representations. Finally we point out related datasets for experiments and future research directions.