Supervised Sentence Fusion with Single-Stage Inference
Kapil Thadani, Kathleen R. McKeown · 2013
Sentence fusion—the merging of sentences containing similar information— has been shown to be useful in an abstractive summarization context. We present a new dataset of sentence fusion instances obtained from evaluation datasets in summarization shared tasks and use this dataset to explore supervised approaches to sentence fusion. Our proposed inference approach recovers the highest scoring output fusion under an n-gram factorization using a compact integer linear programming formulation that avoids cycles and disconnected structures. In addition, we introduce simple fusion-specific features and constraints that outperform a compression-inspired baseline as well as a variant that relies on human-identified concept spans for perfect content selection. 1