# To Interpolate or not to Interpolate: PRF, Dense and Sparse Retrievers

arXiv: https://arxiv.org/abs/2205.00235 · HTML: https://mlsanity.com/paper/2205.00235/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 30 Apr 2022.

## Built on by (most cited first, 12 of 12)

- [FeB4RAG: Evaluating Federated Search in the Context of Retrieval Augmented Generation](https://mlsanity.com/paper/2402.11891/) (2024)
- [LED: Lexicon-Enlightened Dense Retriever for Large-Scale Retrieval](https://mlsanity.com/paper/2208.13661/) (2022)
- [Neural Rankers for Effective Screening Prioritisation in Medical Systematic Review Literature Search](https://mlsanity.com/paper/2212.09017/) (2022)
- [Automated MeSH Term Suggestion for Effective Query Formulation in Systematic Reviews Literature Search](https://mlsanity.com/paper/2209.08687/) (2022)
- [Explain like I am BM25: Interpreting a Dense Model's Ranked-List with a Sparse Approximation](https://mlsanity.com/paper/2304.12631/) (2023)
- [Simple Yet Effective Neural Ranking and Reranking Baselines for Cross-Lingual Information Retrieval](https://mlsanity.com/paper/2304.01019/) (2023)
- [Pseudo Relevance Feedback is Enough to Close the Gap Between Small and Large Dense Retrieval Models](https://mlsanity.com/paper/2503.14887/) (2025)
- [ReFIT: Relevance Feedback from a Reranker during Inference](https://mlsanity.com/paper/2305.11744/) (2023)
- [MICO: Selective Search with Mutual Information Co-training](https://mlsanity.com/paper/2209.04378/) (2022)
- [DiffRetriever: Parallel Representative Tokens for Retrieval with Diffusion Language Models](https://mlsanity.com/paper/2605.07210/) (2026)
- [Towards Consistency Filtering-Free Unsupervised Learning for Dense Retrieval](https://mlsanity.com/paper/2308.02926/) (2023)
- [What's the Best Way to Retrieve Slides? A Comparative Study of Multimodal, Caption-Based, and Hybrid Retrieval Techniques](https://mlsanity.com/paper/2509.15211/) (2025)

## Abstract (arXiv)

Current pre-trained language model approaches to information retrieval can be broadly divided into two categories: sparse retrievers (to which belong also non-neural approaches such as bag-of-words methods, e.g., BM25) and dense retrievers. Each of these categories appears to capture different characteristics of relevance. Previous work has investigated how relevance signals from sparse retrievers could be combined with those from dense retrievers via interpolation. Such interpolation would generally lead to higher retrieval effectiveness. In this paper we consider the problem of combining the relevance signals from sparse and dense retrievers in the context of Pseudo Relevance Feedback (PRF). This context poses two key challenges: (1) When should interpolation occur: before, after, or both before and after the PRF process? (2) Which sparse representation should be considered: a zero-shot bag-of-words model (BM25), or a learnt sparse representation? To answer these questions we perform a thorough empirical evaluation considering an effective and scalable neural PRF approach (Vector-PRF), three effective dense retrievers (ANCE, TCTv2, DistillBERT), and one state-of-the-art learnt sparse retriever (uniCOIL). The empirical findings from our experiments suggest that, regardless of sparse representation and dense retriever, interpolation both before and after PRF achieves the highest effectiveness across most datasets and metrics.