Search papers, labs, and topics across Lattice.
This paper investigates the integration of generated query expansions in dense retrieval systems, revealing that the method of incorporation significantly impacts performance. By introducing AnchorQE, a training-free approach that separately encodes the original query and its expansion, the authors demonstrate that effective interpolation of these components can enhance retrieval effectiveness. Their results show improvements of up to 12.89% over traditional methods across multiple benchmark datasets, highlighting the importance of integration strategies in retrieval tasks.
Query expansions can backfire if not integrated properly, but AnchorQE shows how to harness them for up to 12.89% better retrieval performance.
Large language models (LLMs) can generate query expansions without task-specific training, yet the same expansions often make a frozen dense retriever worse. We identify an underexplored factor: prior work has often focused on what text is generated, while how generated text is incorporated into dense retrievers has received less systematic attention. By holding generated expansions fixed, we show that performance degradation can often be attributed to the integration method itself. We introduce AnchorQE, a training-free method that separately encodes the original query and its expansion before interpolating them. The interpolation factor is estimated using an unsupervised online strategy that operates over a small part of the unlabeled test stream. Intuitively, our strategy assigns high expansion trust only when expansions are both retrieval-strong and consistent with the original query's retrieved evidence. We show that AnchorQE improves retrieval effectiveness by up to 12.89% when compared to widely-used expansion-only or text-level concatenation baselines across TREC-DL, LoTTE, and BEIR. Further, we show that our online strategy to estimate the interpolation factor outperforms a fixed weight tuned on a development partition by up to 3.81%.
