FRUIT: Faithfully Reflecting Updated Information in Text

TL;DR

This paper introduces the FRUIT task and dataset for updating text with new evidence, demonstrating the need for models that can faithfully incorporate external knowledge into existing articles.

Contribution

The paper presents the novel FRUIT task, releases a large dataset, and proposes EDIT5, a new T5-based model that sets a new state of the art for faithful text updating.

Findings

EDIT5 outperforms existing models on the FRUIT benchmark.

Developing models for faithful updates requires new neural generation capabilities.

The FRUIT dataset enables research on knowledge-aware text editing.

Abstract

Textual knowledge bases such as Wikipedia require considerable effort to keep up to date and consistent. While automated writing assistants could potentially ease this burden, the problem of suggesting edits grounded in external knowledge has been under-explored. In this paper, we introduce the novel generation task of *faithfully reflecting updated information in text* (FRUIT) where the goal is to update an existing article given new evidence. We release the FRUIT-WIKI dataset, a collection of over 170K distantly supervised data produced from pairs of Wikipedia snapshots, along with our data generation pipeline and a gold evaluation set of 914 instances whose edits are guaranteed to be supported by the evidence. We provide benchmark results for popular generation systems as well as EDIT5 -- a T5-based approach tailored to editing we introduce that establishes the state of the art. Our analysis shows that developing models that can update articles faithfully requires new capabilities for neural generation models, and opens doors to many new applications.