Retrieve, Rerank and Rewrite: Soft Template Based Neural Summarization

• DOI: 10.18653/v1/p18-1015
• Publication Date: 2019-06-29T15:52:01Z
• Authors: Ziqiang Cao; Wenjie Li; Sujian Li; Furu Wei
• Keywords: Template

Most previous seq2seq summarization systems purely depend on the source text to generate summaries, which tends work unstably. Inspired by traditional template-based approaches, this paper proposes use existing summaries as soft templates guide model. To end, we a popular IR platform Retrieve proper candidate templates. Then, extend framework jointly conduct template Reranking and template-aware summary generation (Rewriting). Experiments show that, in terms of informativeness, our model significantly outperforms state-of-the-art methods, even themselves demonstrate high competitiveness. In addition, import high-quality external improves stability readability generated summaries.