Abstract Meaning Representation for Multi-Document Summarization

TL;DR

This paper explores using Abstract Meaning Representation (AMR) graphs for multi-document summarization, demonstrating a flexible, data-driven approach that generates summaries by transforming semantic graphs into sentences, with promising experimental results.

Contribution

It introduces a novel AMR-based framework for multi-document summarization that is fully data-driven and allows independent optimization of components.

Findings

Promising results on benchmark datasets

Framework is flexible and can be optimized independently

Highlights opportunities and challenges for future research

Abstract

Generating an abstract from a collection of documents is a desirable capability for many real-world applications. However, abstractive approaches to multi-document summarization have not been thoroughly investigated. This paper studies the feasibility of using Abstract Meaning Representation (AMR), a semantic representation of natural language grounded in linguistic theory, as a form of content representation. Our approach condenses source documents to a set of summary graphs following the AMR formalism. The summary graphs are then transformed to a set of summary sentences in a surface realization step. The framework is fully data-driven and flexible. Each component can be optimized independently using small-scale, in-domain training data. We perform experiments on benchmark summarization datasets and report promising results. We also describe opportunities and challenges for advancing this line of research.