BiCA: Effective Biomedical Dense Retrieval with Citation-Aware Hard Negatives
By: Aarush Sinha , Pavan Kumar S , Roshan Balaji and more
Potential Business Impact:
Helps computers find science papers better.
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.
Similar Papers
RRRA: Resampling and Reranking through a Retriever Adapter
Information Retrieval
Teaches computers to find better answers faster.
Enhancing Document Retrieval for Curating N-ary Relations in Knowledge Bases
Information Retrieval
Finds medical facts to build knowledge faster.
Improving Document Retrieval Coherence for Semantically Equivalent Queries
Information Retrieval
Finds better answers even with different questions.