GGP: A Graph-based Grouping Planner for Explicit Control of Long Text Generation

TL;DR

This paper introduces GGP, a graph-based planner that improves long text generation by explicitly controlling content organization and coherence, addressing the limitations of existing short-text focused methods.

Contribution

The paper proposes a novel graph-based grouping planner (GGP) that encodes key phrases into dual representations and regroups them for controlled long text generation.

Findings

GGP significantly outperforms baselines on three datasets.

GGP effectively controls content order and coherence.

GGP demonstrates improved logical consistency in generated texts.

Abstract

Existing data-driven methods can well handle short text generation. However, when applied to the long-text generation scenarios such as story generation or advertising text generation in the commercial scenario, these methods may generate illogical and uncontrollable texts. To address these aforementioned issues, we propose a graph-based grouping planner(GGP) following the idea of first-plan-then-generate. Specifically, given a collection of key phrases, GGP firstly encodes these phrases into an instance-level sequential representation and a corpus-level graph-based representation separately. With these two synergic representations, we then regroup these phrases into a fine-grained plan, based on which we generate the final long text. We conduct our experiments on three long text generation datasets and the experimental results reveal that GGP significantly outperforms baselines, which proves that GGP can control the long text generation by knowing how to say and in what order.