# Plan ahead: Self-Supervised Text Planning for Paragraph Completion Task

arXiv: https://arxiv.org/abs/2010.05141 · HTML: https://mlsanity.com/paper/2010.05141/

## Key facts

- Cited by 13 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 11 Oct 2020.

## Built on by (most cited first, 13 of 13)

- [LongGenBench: Benchmarking Long-Form Generation in Long Context LLMs](https://mlsanity.com/paper/2409.02076/) (2024)
- [HelloBench: Evaluating Long Text Generation Capabilities of Large Language Models](https://mlsanity.com/paper/2409.16191/) (2024)
- [ORGAN: Observation-Guided Radiology Report Generation via Tree Reasoning](https://mlsanity.com/paper/2306.06466/) (2023)
- [Spatial-Temporal Hypergraph Self-Supervised Learning for Crime Prediction](https://mlsanity.com/paper/2204.08587/) (2022)
- [PLANET: Dynamic Content Planning in Autoregressive Transformers for Long-form Text Generation](https://mlsanity.com/paper/2203.09100/) (2022)
- [PROXYQA: An Alternative Framework for Evaluating Long-Form Text Generation with Large Language Models](https://mlsanity.com/paper/2401.15042/) (2024)
- [Stylized Story Generation with Style-Guided Planning](https://mlsanity.com/paper/2105.08625/) (2021)
- [Persona-Guided Planning for Controlling the Protagonist's Persona in Story Generation](https://mlsanity.com/paper/2204.10703/) (2022)
- [AMERICANO: Argument Generation with Discourse-driven Decomposition and Agent Interaction](https://mlsanity.com/paper/2310.20352/) (2023)
- [Generating Coherent Narratives by Learning Dynamic and Discrete Entity States with a Contrastive Framework](https://mlsanity.com/paper/2208.03985/) (2022)
- [EASE: Extractive-Abstractive Summarization with Explanations](https://mlsanity.com/paper/2105.06982/) (2021)
- [MOCHA: A Multi-Task Training Approach for Coherent Text Generation from Cognitive Perspective](https://mlsanity.com/paper/2210.14650/) (2022)
- [Integrating AI Planning with Natural Language Processing: A Combination of Explicit and Tacit Knowledge](https://mlsanity.com/paper/2202.07138/) (2022)

## Abstract (arXiv)

Despite the recent success of contextualized language models on various NLP tasks, language model itself cannot capture textual coherence of a long, multi-sentence document (e.g., a paragraph). Humans often make structural decisions on what and how to say about before making utterances. Guiding surface realization with such high-level decisions and structuring text in a coherent way is essentially called a planning process. Where can the model learn such high-level coherence? A paragraph itself contains various forms of inductive coherence signals called self-supervision in this work, such as sentence orders, topical keywords, rhetorical structures, and so on. Motivated by that, this work proposes a new paragraph completion task PARCOM; predicting masked sentences in a paragraph. However, the task suffers from predicting and selecting appropriate topical content with respect to the given context. To address that, we propose a self-supervised text planner SSPlanner that predicts what to say first (content prediction), then guides the pretrained language model (surface realization) using the predicted content. SSPlanner outperforms the baseline generation models on the paragraph completion task in both automatic and human evaluation. We also find that a combination of noun and verb types of keywords is the most effective for content selection. As more number of content keywords are provided, overall generation quality also increases.