Restructuring Conversations using Discourse Relations for Zero-shot Abstractive Dialogue Summarization
This addresses the challenge of summarizing informal dialogues for applications like meeting analysis, offering a zero-shot solution that reduces reliance on annotated data, though it is incremental as it builds on existing summarization models.
The authors tackled the problem of abstractive dialogue summarization by restructuring conversations using discourse relations, enabling zero-shot summarization without domain-specific training data. Their method improved ROGUE scores by up to 3 points on meeting corpora and performed competitively against state-of-the-art approaches.
Dialogue summarization is a challenging problem due to the informal and unstructured nature of conversational data. Recent advances in abstractive summarization have been focused on data-hungry neural models and adapting these models to a new domain requires the availability of domain-specific manually annotated corpus created by linguistic experts. We propose a zero-shot abstractive dialogue summarization method that uses discourse relations to provide structure to conversations, and then uses an out-of-the-box document summarization model to create final summaries. Experiments on the AMI and ICSI meeting corpus, with document summarization models like PGN and BART, shows that our method improves the ROGUE score by up to 3 points, and even performs competitively against other state-of-the-art methods.