Post-Training Dialogue Summarization using Pseudo-Paraphrasing
This work addresses dialogue summarization for natural language processing applications, offering a more efficient and effective method compared to existing approaches.
The paper tackles the format gap between dialogues and narrative summaries in dialogue summarization by post-training pretrained language models to rephrase dialogues into narratives before fine-tuning, resulting in significant improvements over vanilla models and outperforming other state-of-the-art models in summary quality and implementation costs.
Previous dialogue summarization techniques adapt large language models pretrained on the narrative text by injecting dialogue-specific features into the models. These features either require additional knowledge to recognize or make the resulting models harder to tune. To bridge the format gap between dialogues and narrative summaries in dialogue summarization tasks, we propose to post-train pretrained language models (PLMs) to rephrase from dialogue to narratives. After that, the model is fine-tuned for dialogue summarization as usual. Comprehensive experiments show that our approach significantly improves vanilla PLMs on dialogue summarization and outperforms other SOTA models by the summary quality and implementation costs.