Effective Unsupervised Constrained Text Generation based on Perturbed Masking

TL;DR

The paper tackles unsupervised constrained text generation by introducing PMCTG, which extends perturbed masking to identify the best edit position and employs multi‑aspect scoring to select the edit action at each step. The approach reduces search steps and improves constraint satisfaction without requiring supervised data, achieving strong performance on keywords-to-sentence generation and paraphrasing. It demonstrates that a principled combination of position probing and action scoring can match or exceed state-of-the-art results in unsupervised settings while preserving flexibility across tasks. This method offers practical value for deploying constrained generation in diverse domains without large parallel corpora.

Abstract

Unsupervised constrained text generation aims to generate text under a given set of constraints without any supervised data. Current state-of-the-art methods stochastically sample edit positions and actions, which may cause unnecessary search steps. In this paper, we propose PMCTG to improve effectiveness by searching for the best edit position and action in each step. Specifically, PMCTG extends perturbed masking technique to effectively search for the most incongruent token to edit. Then it introduces four multi-aspect scoring functions to select edit action to further reduce search difficulty. Since PMCTG does not require supervised data, it could be applied to different generation tasks. We show that under the unsupervised setting, PMCTG achieves new state-of-the-art results in two representative tasks, namely keywords-to-sentence generation and paraphrasing.