HiPrompt: Few-Shot Biomedical Knowledge Fusion via Hierarchy-Oriented Prompting
This work addresses the challenge of fusing biomedical knowledge graphs for medical decision-making, offering a novel approach that reduces reliance on labeled data, though it is incremental in leveraging existing LLM capabilities.
The paper tackles the problem of biomedical knowledge fusion with scarce supervision by proposing HiPrompt, a framework that uses hierarchy-oriented prompting with large language models for few-shot reasoning, achieving effective results on benchmark datasets.
Medical decision-making processes can be enhanced by comprehensive biomedical knowledge bases, which require fusing knowledge graphs constructed from different sources via a uniform index system. The index system often organizes biomedical terms in a hierarchy to provide the aligned entities with fine-grained granularity. To address the challenge of scarce supervision in the biomedical knowledge fusion (BKF) task, researchers have proposed various unsupervised methods. However, these methods heavily rely on ad-hoc lexical and structural matching algorithms, which fail to capture the rich semantics conveyed by biomedical entities and terms. Recently, neural embedding models have proved effective in semantic-rich tasks, but they rely on sufficient labeled data to be adequately trained. To bridge the gap between the scarce-labeled BKF and neural embedding models, we propose HiPrompt, a supervision-efficient knowledge fusion framework that elicits the few-shot reasoning ability of large language models through hierarchy-oriented prompts. Empirical results on the collected KG-Hi-BKF benchmark datasets demonstrate the effectiveness of HiPrompt.