Learning Mixtures of Submodular Shells with Application to Document Summarization

TL;DR

This paper presents a novel method for learning mixtures of submodular functions called shells, optimized with large-margin structured prediction, and demonstrates its effectiveness in multi-document summarization, achieving state-of-the-art results.

Contribution

It introduces a new approach to learn mixtures of submodular shells with theoretical guarantees, applied successfully to document summarization.

Findings

Achieved the best results on NIST DUC-05 to DUC-07 datasets.

Provided a risk bound guarantee for large-margin learning with approximate submodular optimization.

Demonstrated the effectiveness of the method in multi-document summarization tasks.

Abstract

We introduce a method to learn a mixture of submodular "shells" in a large-margin setting. A submodular shell is an abstract submodular function that can be instantiated with a ground set and a set of parameters to produce a submodular function. A mixture of such shells can then also be so instantiated to produce a more complex submodular function. What our algorithm learns are the mixture weights over such shells. We provide a risk bound guarantee when learning in a large-margin structured-prediction setting using a projected subgradient method when only approximate submodular optimization is possible (such as with submodular function maximization). We apply this method to the problem of multi-document summarization and produce the best results reported so far on the widely used NIST DUC-05 through DUC-07 document summarization corpora.