Domain-Oriented Prefix-Tuning: Towards Efficient and Generalizable Fine-tuning for Zero-Shot Dialogue Summarization
This work addresses the lack of generalization in dialogue summarization models for new domains, offering a lightweight fine-tuning method that is incremental over existing prefix-tuning approaches.
The paper tackles the problem of domain adaptation in zero-shot dialogue summarization by proposing Domain-Oriented Prefix-Tuning, which uses domain word initialization and discrete prompts to improve generalization, achieving effectiveness as proven by experiments on TODSum and QMSum datasets.
The most advanced abstractive dialogue summarizers lack generalization ability on new domains and the existing researches for domain adaptation in summarization generally rely on large-scale pre-trainings. To explore the lightweight fine-tuning methods for domain adaptation of dialogue summarization, in this paper, we propose an efficient and generalizable Domain-Oriented Prefix-tuning model, which utilizes a domain word initialized prefix module to alleviate domain entanglement and adopts discrete prompts to guide the model to focus on key contents of dialogues and enhance model generalization. We conduct zero-shot experiments and build domain adaptation benchmarks on two multi-domain dialogue summarization datasets, TODSum and QMSum. Adequate experiments and qualitative analysis prove the effectiveness of our methods.