# DIFFQG: Generating Questions to Summarize Factual Changes

## TL;DR

This paper introduces DIFFQG, a dataset of question-answer pairs that effectively capture factual changes between paired documents, aiding automatic update summarization and understanding document evolution.

## Contribution

The paper presents DIFFQG, a novel dataset of QA pairs for factual change detection, and demonstrates its utility in summarizing document updates beyond existing methods.

## Key findings

- DIFFQG contains 759 QA pairs and 1153 non-changing examples.
- Questions in DIFFQG are unambiguous, complex, and information-seeking.
- The dataset advances automatic update summarization and factual change detection.

## Abstract

Identifying the difference between two versions of the same article is useful to update knowledge bases and to understand how articles evolve. Paired texts occur naturally in diverse situations: reporters write similar news stories and maintainers of authoritative websites must keep their information up to date. We propose representing factual changes between paired documents as question-answer pairs, where the answer to the same question differs between two versions. We find that question-answer pairs can flexibly and concisely capture the updated contents. Provided with paired documents, annotators identify questions that are answered by one passage but answered differently or cannot be answered by the other. We release DIFFQG which consists of 759 QA pairs and 1153 examples of paired passages with no factual change. These questions are intended to be both unambiguous and information-seeking and involve complex edits, pushing beyond the capabilities of current question generation and factual change detection systems. Our dataset summarizes the changes between two versions of the document as questions and answers, studying automatic update summarization in a novel way.