Towards Coherent Multi-Document Summarization

Janara Christensen, Mausam Mausam, Stephen Soderland and Oren Etzioni

This paper presents G-FLOW, a novel system for coherent extractive multi-document summarization (MDS). Where previous work on MDS considered sentence selection and ordering separately, G-FLOW introduces a joint model for selection and ordering that balances coherence and salience. G-FLOW’s core representation is a graph that approximates the discourse relations across sentences based on indicators including discourse cues, deverbal nouns, co-reference, and more. This graph enables G-FLOW to estimate the coherence of a candidate summary. We evaluate G-FLOW on Mechanical Turk, and find that it generates dramatically better summaries than an extractive summarizer based on a pipeline of state-of-the-art sentence selection and reordering components, underscoring the value of our joint model.