BiCA: 基于引用感知的生物医学密集检索
BiCA: Effective Biomedical Dense Retrieval with Citation-Aware Hard Negatives
- University of Copenhagen（哥本哈根大学）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
BiCA通过利用引用链接生成引用感知硬负样本，提升生物医学领域密集检索性能，实现零样本检索和长尾主题的优越表现。
AI中文摘要：
硬负样本是训练有效检索模型的关键。硬负样本挖掘通常依赖于使用交叉编码器或静态嵌入模型根据相似度度量（如余弦距离）对文档进行排序。然而，在生物医学和科学领域，由于难以区分源文档和硬负样本文档，硬负样本挖掘变得具有挑战性。然而，引用文档自然与源文档共享上下文相关性，但并非重复，因此非常适合用作硬负样本。在本文中，我们提出BiCA：具有引用感知硬负样本的生物医学密集检索，通过利用20000篇PubMed文章中的引用链接来改进一个领域特定的小型密集检索器。我们使用这些引用信息的负样本对GTE_small和GTE_Base模型进行微调，并观察到在BEIR上进行域内和域外任务时，使用nDCG@10的零样本密集检索性能有持续提升，并且在LoTTE上的长尾主题中，使用Success@5优于基线。我们的发现强调了利用文档链接结构生成高信息量负样本的潜力，使在最小微调的情况下达到最先进的性能，并展示了通往高数据效率领域适应的路径。
英文摘要：
Hard negatives are essential for training effective retrieval models. Hard-negative mining typically relies on ranking documents using cross-encoders or static embedding models based on similarity metrics such as cosine distance. Hard negative mining becomes challenging for biomedical and scientific domains due to the difficulty in distinguishing between source and hard negative documents. However, referenced documents naturally share contextual relevance with the source document but are not duplicates, making them well-suited as hard negatives. In this work, we propose BiCA: Biomedical Dense Retrieval with Citation-Aware Hard Negatives, an approach for hard-negative mining by utilizing citation links in 20,000 PubMed articles for improving a domain-specific small dense retriever. We fine-tune the GTE_small and GTE_Base models using these citation-informed negatives and observe consistent improvements in zero-shot dense retrieval using nDCG@10 for both in-domain and out-of-domain tasks on BEIR and outperform baselines on long-tailed topics in LoTTE using Success@5. Our findings highlight the potential of leveraging document link structure to generate highly informative negatives, enabling state-of-the-art performance with minimal fine-tuning and demonstrating a path towards highly data-efficient domain adaptation.
