Where did you get that? Towards Summarization Attribution for Analysts

TL;DR

This paper investigates automatic attribution methods for summaries, linking sentences to source texts, and proposes a hybrid summarization approach to improve attribution accuracy, along with analyzing attribution errors.

Contribution

It introduces a hybrid summarization technique for better attribution and a custom topology to categorize attribution errors, advancing automatic attribution methods.

Findings

Hybrid summarization aids attribution clarity

Custom topology effectively categorizes attribution errors

Proposed methods improve attribution accuracy

Abstract

Analysts require attribution, as nothing can be reported without knowing the source of the information. In this paper, we will focus on automatic methods for attribution, linking each sentence in the summary to a portion of the source text, which may be in one or more documents. We explore using a hybrid summarization, i.e., an automatic paraphrase of an extractive summary, to ease attribution. We also use a custom topology to identify the proportion of different categories of attribution-related errors.