PLANET: Dynamic Content Planning in Autoregressive Transformers for Long-form Text Generation

TL;DR

PLANET introduces a dynamic content planning framework for long-form text generation using autoregressive Transformers, significantly improving coherence and content richness through latent semantic planning and contrastive learning.

Contribution

The paper presents a novel framework that integrates dynamic content planning into Transformer-based models, enhancing coherence and content control in long-form text generation.

Findings

Outperforms strong baselines in coherence and content richness.

Improves long-form text coherence via contrastive learning.

Effective in counterargument and opinion article generation.

Abstract

Despite recent progress of pre-trained language models on generating fluent text, existing methods still suffer from incoherence problems in long-form text generation tasks that require proper content control and planning to form a coherent high-level logical flow. In this work, we propose PLANET, a novel generation framework leveraging autoregressive self-attention mechanism to conduct content planning and surface realization dynamically. To guide the generation of output sentences, our framework enriches the Transformer decoder with latent representations to maintain sentence-level semantic plans grounded by bag-of-words. Moreover, we introduce a new coherence-based contrastive learning objective to further improve the coherence of output. Extensive experiments are conducted on two challenging long-form text generation tasks including counterargument generation and opinion article generation. Both automatic and human evaluations show that our method significantly outperforms strong baselines and generates more coherent texts with richer contents.