Generating Summaries with Topic Templates and Structured Convolutional Decoders
This addresses the issue of content coverage in text summarization for NLP applications, representing an incremental improvement.
The paper tackled the problem of generating multi-sentence summaries by proposing a structured convolutional decoder guided by content structure, resulting in summaries with better content coverage compared to existing sequential decoders across three datasets.
Existing neural generation approaches create multi-sentence text as a single sequence. In this paper we propose a structured convolutional decoder that is guided by the content structure of target summaries. We compare our model with existing sequential decoders on three data sets representing different domains. Automatic and human evaluation demonstrate that our summaries have better content coverage.