A Bag of Tricks for Dialogue Summarization
This work addresses dialogue summarization for applications like meeting notes or customer service, but it is incremental as it builds on existing pretrained models with specific tweaks.
The paper tackled dialogue summarization by addressing challenges like speaker differentiation, negation, reasoning, and informal language, using techniques such as speaker name substitution and multi-task learning, resulting in improved performance that outperformed strong baselines.
Dialogue summarization comes with its own peculiar challenges as opposed to news or scientific articles summarization. In this work, we explore four different challenges of the task: handling and differentiating parts of the dialogue belonging to multiple speakers, negation understanding, reasoning about the situation, and informal language understanding. Using a pretrained sequence-to-sequence language model, we explore speaker name substitution, negation scope highlighting, multi-task learning with relevant tasks, and pretraining on in-domain data. Our experiments show that our proposed techniques indeed improve summarization performance, outperforming strong baselines.