Counterfactual Story Reasoning and Generation

TL;DR

This paper introduces a new task and dataset for counterfactual story rewriting, requiring models to minimally revise narratives to incorporate alternative events, advancing understanding of causal reasoning in language models.

Contribution

It proposes the task of counterfactual story rewriting, creates the TimeTravel dataset with nearly 30,000 examples, and evaluates baseline models' capabilities in this challenging reasoning task.

Findings

Baseline models show limited success in human-aligned counterfactual rewriting.

Automatic metrics often do not correlate well with human judgments.

The dataset enables future research on causal reasoning in narrative generation.

Abstract

Counterfactual reasoning requires predicting how alternative events, contrary to what actually happened, might have resulted in different outcomes. Despite being considered a necessary component of AI-complete systems, few resources have been developed for evaluating counterfactual reasoning in narratives. In this paper, we propose Counterfactual Story Rewriting: given an original story and an intervening counterfactual event, the task is to minimally revise the story to make it compatible with the given counterfactual event. Solving this task will require deep understanding of causal narrative chains and counterfactual invariance, and integration of such story reasoning capabilities into conditional language generation models. We present TimeTravel, a new dataset of 29,849 counterfactual rewritings, each with the original story, a counterfactual event, and human-generated revision of the original story compatible with the counterfactual event. Additionally, we include 80,115 counterfactual "branches" without a rewritten storyline to support future work on semi- or un-supervised approaches to counterfactual story rewriting. Finally, we evaluate the counterfactual rewriting capacities of several competitive baselines based on pretrained language models, and assess whether common overlap and model-based automatic metrics for text generation correlate well with human scores for counterfactual rewriting.