INSET: Sentence Infilling with INter-SEntential Transformer

TL;DR

This paper introduces INSET, a transformer-based framework that improves sentence infilling by decoupling understanding, planning, and generation, leveraging pre-trained models like BERT and GPT-2.

Contribution

It proposes a novel framework that separates the challenges of understanding, planning, and generation in sentence infilling, utilizing existing large-scale pre-trained models.

Findings

Effective learning of sentence representations for generation

Successful generation of contextually fitting missing sentences

Demonstrated improvements over baseline models

Abstract

Missing sentence generation (or sentence infilling) fosters a wide range of applications in natural language generation, such as document auto-completion and meeting note expansion. This task asks the model to generate intermediate missing sentences that can syntactically and semantically bridge the surrounding context. Solving the sentence infilling task requires techniques in natural language processing ranging from understanding to discourse-level planning to generation. In this paper, we propose a framework to decouple the challenge and address these three aspects respectively, leveraging the power of existing large-scale pre-trained models such as BERT and GPT-2. We empirically demonstrate the effectiveness of our model in learning a sentence representation for generation and further generating a missing sentence that fits the context.