Generating Coherent Narratives by Learning Dynamic and Discrete Entity States with a Contrastive Framework

TL;DR

This paper introduces a novel narrative generation method that models dynamic entity states using a contrastive learning framework, resulting in more coherent and diverse stories compared to existing models.

Contribution

It extends the Transformer with dynamic entity state updates and a contrastive learning approach to improve narrative coherence and diversity.

Findings

Generated narratives are more coherent and diverse.

The model outperforms strong baselines on two datasets.

Entity state representations are effectively learned in a discrete space.

Abstract

Despite advances in generating fluent texts, existing pretraining models tend to attach incoherent event sequences to involved entities when generating narratives such as stories and news. We conjecture that such issues result from representing entities as static embeddings of superficial words, while neglecting to model their ever-changing states, i.e., the information they carry, as the text unfolds. Therefore, we extend the Transformer model to dynamically conduct entity state updates and sentence realization for narrative generation. We propose a contrastive framework to learn the state representations in a discrete space, and insert additional attention layers into the decoder to better exploit these states. Experiments on two narrative datasets show that our model can generate more coherent and diverse narratives than strong baselines with the guidance of meaningful entity states.