tsta

Home / دومین کنفرانس ملی عصر انفجار تکنولوژی؛ هوش مصنوعی، تحولی در صنعت، تجارت و زنجیره تامین و دومین کنفرانس ملی علم داده در کاربردهای مهندسی

A Multi-Stage Ranking Pipeline for High-Precision Medical Information Retrieval

Authors :

Asa Shabanian¹ Alireza Asl Nemati² Morteza Mohammadi Zanjireh³

1- دانشگاه بین الملل امام خمینی 2- دانشگاه بین الملل امام خمینی 3- دانشگاه بین الملل امام خمینی

Keywords :

Information Retrieval،Hybrid Search،Re-ranking،Large Language Models (LLMs)

Abstract :

Effective retrieval of biomedical information presents a significant challenge due to terminological complexity and semantic ambiguity. Traditional keyword-based methods like BM25 often fail to capture the user's semantic intent. To address this, we propose and empirically evaluate a multi-stage ranking architecture designed for high-precision retrieval. Our pipeline initiates with two parallel retrieval stages: a sparse lexical retriever (BM25) and a dense semantic retriever using a Bi-Encoder model (multi-qa-MiniLM-L6-cos-v1). The resulting candidate lists are then fused using Reciprocal Rank Fusion (RRF) to leverage their complementary strengths. In the final stage, a more powerful Cross-Encoder model (ms-marco-MiniLM-L-6-v2) re-ranks the top 100 candidates from the fused list to achieve fine-grained relevance scoring. Evaluated on the standard TREC-COVID dataset, our complete pipeline demonstrates substantial performance gains at each stage, culminating in a final Precision@10 of 0.808 and an nDCG@10 of 0.754. This represents a significant relative improvement of 68% and 69%, respectively, over the BM25 baseline. These results validate the efficacy of a cascaded retrieve-fuse-rerank architecture. Our work underscores the synergistic value of combining sparse, dense, and cross-attention models, providing a robust framework for developing high-performance information retrieval systems in specialized domains.