TopClustRAG at SIGIR 2025 LiveRAG Challenge

TL;DR

TopClustRAG is a hybrid retrieval-augmented generation system that uses clustering and multi-stage filtering to improve answer quality in large-scale web question answering tasks.

Contribution

It introduces a novel clustering-based approach for context filtering and prompt aggregation in RAG systems, enhancing answer relevance and faithfulness.

Findings

Ranked 2nd in faithfulness on the leaderboard

Achieved 7th in correctness on the leaderboard

Demonstrated effectiveness of clustering in large-scale RAG

Abstract

We present TopClustRAG, a retrieval-augmented generation (RAG) system developed for the LiveRAG Challenge, which evaluates end-to-end question answering over large-scale web corpora. Our system employs a hybrid retrieval strategy combining sparse and dense indices, followed by K-Means clustering to group semantically similar passages. Representative passages from each cluster are used to construct cluster-specific prompts for a large language model (LLM), generating intermediate answers that are filtered, reranked, and finally synthesized into a single, comprehensive response. This multi-stage pipeline enhances answer diversity, relevance, and faithfulness to retrieved evidence. Evaluated on the FineWeb Sample-10BT dataset, TopClustRAG ranked 2nd in faithfulness and 7th in correctness on the official leaderboard, demonstrating the effectiveness of clustering-based context filtering and prompt aggregation in large-scale RAG systems.