ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT
The DOI matches its Crossref record; 26 of the 26 references it cites were found in a registry or an index.
Abstract
Recent progress in Natural Language Understanding (NLU) is driving fast-paced advances in Information Retrieval (IR), largely owed to fine-tuning deep language models (LMs) for document ranking. While remarkably effective, the ranking models based on these LMs increase computational cost by orders of magnitude over prior approaches, particularly as they must feed each query-document pair through a massive neural network to compute a single relevance score. To tackle this, we present ColBERT, a novel ranking model that adapts deep LMs (in particular, BERT) for efficient retrieval. ColBERT introduces a late interaction architecture that independently encodes the query and the document using BERT and then employs a cheap yet powerful interaction step that models their fine-grained similarity. By delaying and yet retaining this fine-granular interaction, ColBERT can leverage the expressiveness of deep LMs while simultaneously gaining the ability to pre-compute document representations offline, considerably speeding up query processing. Crucially, ColBERT's pruning-friendly interaction mechanism enables leveraging vector-similarity indexes for end-to-end retrieval directly from millions of documents. We extensively evaluate ColBERT using two recent passage search datasets. Results show that ColBERT's effectiveness is competitive with existing BERT-based models (and outperforms every non-BERT baseline), while executing two orders-of-magnitude faster and requiring up to four orders-of-magnitude fewer FLOPs per query.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“The only ML component is a CrossEncoder reranker ( mxbai-rerank-large-v1 , 435M parameters, DeBERTaV3) fused with ColBERT (Khattab & Zaharia, 2020) via Reciprocal Rank Fusion (RRF), applied after retrieval is complete to prioritize passages before token-budget truncation.”
From SmartSearch: How Ranking Beats Structure for Conversational Memory Retrieval (2026) · cites this paper as Khattab & Zaharia, 2020
In the citing paper
“While dense retrievers like Dense Passage Retrieval (DPR) [14] and late-interaction models like ColBERT [15] improve recall on single-hop queries, they often struggle with multi-hop reasoning, where the answer depends on composing information from multiple disjoint documents [20, 6].”
From Replace, Don't Expand: Mitigating Context Dilution in Multi-Hop RAG via Fixed-Budget Evidence Assembly (2025) · cites this paper as [15]
In the citing paper
“Research on retrieval optimization has explored multiple techniques, yet studies specifically focused on prefetching size in RaLM systems are still limited.”
From Exploring Optimal Prefetching Sizes in RaLMSpec to Enhance Retrieval-Augmented Generation Efficiency (2025) · cites this paper as [9]
In the citing paper
“Early RAG implementations, however, often relied on document retrieval methods that introduced noise and excessive context [26, 55, 60, 122].”
From LEGO-GraphRAG: Modularizing Graph-Based Retrieval-Augmented Generation for Design Space Exploration (2025) · cites this paper as [55]
In the citing paper
“Standard retrieval methods like RAG and MemoryBank (Zhong et al., 2024) rely fundamentally on vector similarity (Karpukhin et al., 2020; Khattab and Zaharia, 2020), representing memories as isolated points in embedding space (Hu et al., 2025). Consequently, they struggle with queries requiring causal bridging between semantically dissimilar or distant events (Yang et al., 2018; Qi et al., 2019; Tr”
From Synapse: Empowering LLM Agents with Episodic-Semantic Memory via Spreading Activation (2026) · cites this paper as Khattab and Zaharia, 2020
Citations
Cites
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017) · arXiv · doi:10.48550/arxiv.1706.03762 · Cited by 120,404 · cited 1×
“At a high level, TK improves the KNRM architecture described earlier: while KNRM employs kernel pooling on top of word-embedding-based interaction, TK uses a Transformer [34] component for contextually encoding queries and documents before kernel pooling.”
MethodBERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2018) · arXiv · doi:10.48550/arxiv.1810.04805 · Cited by 73,173 · cited 2×
“Among them, a recent approach has emerged that ne-tunes deep pre-trained language models (LMs) like ELMo [29] and BERT [5] for estimating relevance.”
BackgroundGoogle's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation
Wu, Y., Schuster, M., Chen, Z., Le, Q. V., Norouzi, M., Macherey, W., Krikun, M., Cao, Y., Gao, Q., Macherey, K., et al. (2016) · arXiv · doi:10.48550/arxiv.1609.08144 · Cited by 5,319
Billion-scale similarity search with GPUs
Johnson, J., Douze, M., & Jégou, H. (2017) · arXiv · doi:10.48550/arxiv.1702.08734 · Cited by 4,140 · cited 2×
“Additionally, it enables ColBERT to leverage vector-similarity search indexes (e.g., [1, 16]) to retrieve the top-k results directly from a large document collection, substantially improving recall over models that only re-rank the output of term-based retrieval.”
MethodMS MARCO: A Human Generated MAchine Reading COmprehension Dataset
Bajaj, P., Campos, D., Craswell, N., Deng, L., Gao, J., Liu, X., Majumder, R., McNamara, A., Mitra, B., Nguyen, T., et al. (2016) · arXiv · doi:10.48550/arxiv.1611.09268 · Cited by 2,473
Robertson, S. E., Walker, S., Jones, S., Hancock‐Beaulieu, M., & Gatford, M. (1994) · Text REtrieval Conference · Cited by 2,212
Learning deep structured semantic models for web search using clickthrough data
Huang, P.-S., He, X., Gao, J., Deng, L., Acero, A., & Heck, L. (2013) · Proceedings of the 22nd ACM international conference on Information & Knowledge Management · doi:10.1145/2505515.2505665 · Cited by 2,009 · cited 1×
“On the le, Figure 2 (a) illustrates representation-focused rankers, which independently compute an embedding for q and another for d and estimate relevance as a single similarity score between two vectors [13, 41].”
BackgroundTinyBERT: Distilling BERT for natural language understanding
Jiao, X., Yin, Y., Shang, L., Jiang, X., Chen, X., Li, L., Wang, F., & Liu, Q. (2020) · Findings of the Association for Computational Linguistics: EMNLP 2020 · doi:10.18653/v1/2020.findings-emnlp.372 · Cited by 1,786
A Deep Relevance Matching Model for Ad-hoc Retrieval
Guo, J., Fan, Y., Ai, Q., & Croft, W. B. (2016) · Proceedings of the 25th ACM International on Conference on Information and Knowledge Management · doi:10.1145/2983323.2983769 · Cited by 806 · cited 1×
“Over the past few years, the Information Retrieval (IR) community has witnessed the introduction of a host of neural ranking models, including DRMM [8], KNRM [5, 36], and Duet [21, 23].”
BackgroundNogueira, R., & Cho, K. (2019) · arXiv · doi:10.48550/arxiv.1901.04085 · Cited by 615 · cited 9×
“Similar to [25], we fine-tune all ColBERT models with learning rate 3 × 10 − 6 with a batch size 32.”
BackgroundMethodInfluential
Research works printed but not linked: 14
- Zhuyun Dai and Jamie Callan . 2019 a. Context-Aware Sentence/Passage Term Importance Estimation For First Stage Retrieval. arXiv preprint arXiv:1910.10687 ( 2019 ). Zhuyun Dai and Jamie Callan. 2019 a. Context-Aware Sentence/Passage Term Importance Estimation For First Stage Retrieval. arXiv preprint arXiv:1910.10687 (2019). · arXiv
- Zhuyun Dai and Jamie Callan . 2019 b. Deeper Text Understanding for IR with Contextual Neural Language Modeling. arXiv preprint arXiv:1905.09217 ( 2019 ). Zhuyun Dai and Jamie Callan. 2019 b. Deeper Text Understanding for IR with Contextual Neural Language Modeling. arXiv preprint arXiv:1905.09217 (2019). · arXiv
- Jiafeng Guo , Yixing Fan , Liang Pang , Liu Yang , Qingyao Ai , Hamed Zamani , Chen Wu , W Bruce Croft , and Xueqi Cheng . 2019. A deep look into neural ranking models for information retrieval. arXiv preprint arXiv:1903.06902 ( 2019 ). Jiafeng Guo, Yixing Fan, Liang Pang, Liu Yang, Qingyao Ai, Hamed Zamani, Chen Wu, W Bruce Croft, and Xueqi Cheng. 2019. A deep look into neural ranking models for information retrieval. arXiv preprint arXiv:1903.06902 (2019). · arXiv
- Sebastian Hofst"atter and Allan Hanbury . 2019. Let's measure run time! Extending the IR replicability infrastructure to include performance aspects. arXiv preprint arXiv:1907.04614 ( 2019 ). Sebastian Hofst"atter and Allan Hanbury. 2019. Let's measure run time! Extending the IR replicability infrastructure to include performance aspects. arXiv preprint arXiv:1907.04614 (2019). · arXiv
- Sebastian Hofst"atter, Markus Zlabinger , and Allan Hanbury . 2019 b. TU Wien@ TREC Deep Learning'19--Simple Contextualization for Re-ranking. arXiv preprint arXiv:1912.01385 ( 2019 ). Sebastian Hofst"atter, Markus Zlabinger, and Allan Hanbury. 2019 b. TU Wien@ TREC Deep Learning'19--Simple Contextualization for Re-ranking. arXiv preprint arXiv:1912.01385 (2019). · arXiv
- Diederik P Kingma and Jimmy Ba . 2014 . Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980 (2014). Diederik P Kingma and Jimmy Ba. 2014. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980 (2014). · arXiv
- Bhaskar Mitra and Nick Craswell . 2019. An Updated Duet Model for Passage Re-ranking. arXiv preprint arXiv:1903.07666 ( 2019 ). Bhaskar Mitra and Nick Craswell. 2019. An Updated Duet Model for Passage Re-ranking. arXiv preprint arXiv:1903.07666 (2019). · arXiv
- Bhaskar Mitra , Corby Rosset , David Hawking , Nick Craswell , Fernando Diaz , and Emine Yilmaz . 2019. Incorporating query term independence assumption for efficient retrieval and ranking using deep neural networks. arXiv preprint arXiv:1907.03693 ( 2019 ). Bhaskar Mitra, Corby Rosset, David Hawking, Nick Craswell, Fernando Diaz, and Emine Yilmaz. 2019. Incorporating query term independence assumption for efficient retrieval and ranking using deep neural networks. arXiv preprint arXiv:1907.03693 (2019). · arXiv
- Rodrigo Nogueira , Wei Yang , Kyunghyun Cho , and Jimmy Lin . 2019 b. Multi-Stage Document Ranking with BERT. arXiv preprint arXiv:1910.14424 ( 2019 ). Rodrigo Nogueira, Wei Yang, Kyunghyun Cho, and Jimmy Lin. 2019 b. Multi-Stage Document Ranking with BERT. arXiv preprint arXiv:1910.14424 (2019). · arXiv
- Rodrigo Nogueira , Wei Yang , Jimmy Lin , and Kyunghyun Cho . 2019 c. Document Expansion by Query Prediction. arXiv preprint arXiv:1904.08375 ( 2019 ). Rodrigo Nogueira, Wei Yang, Jimmy Lin, and Kyunghyun Cho. 2019 c. Document Expansion by Query Prediction. arXiv preprint arXiv:1904.08375 (2019). · arXiv
- Matthew E Peters , Mark Neumann , Mohit Iyyer , Matt Gardner , Christopher Clark , Kenton Lee , and Luke Zettlemoyer . 2018. Deep contextualized word representations. arXiv preprint arXiv:1802.05365 ( 2018 ). Matthew E Peters, Mark Neumann, Mohit Iyyer, Matt Gardner, Christopher Clark, Kenton Lee, and Luke Zettlemoyer. 2018. Deep contextualized word representations. arXiv preprint arXiv:1802.05365 (2018). · arXiv
- Yifan Qiao , Chenyan Xiong , Zhenghao Liu , and Zhiyuan Liu . 2019. Understanding the Behaviors of BERT in Ranking. arXiv preprint arXiv:1904.07531 ( 2019 ). Yifan Qiao, Chenyan Xiong, Zhenghao Liu, and Zhiyuan Liu. 2019. Understanding the Behaviors of BERT in Ranking. arXiv preprint arXiv:1904.07531 (2019). · arXiv
- Colin Raffel , Noam Shazeer , Adam Roberts , Katherine Lee , Sharan Narang , Michael Matena , Yanqi Zhou , Wei Li , and Peter J Liu . 2019. Exploring the limits of transfer learning with a unified text-to-text transformer. arXiv preprint arXiv:1910.10683 ( 2019 ). Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J Liu. 2019. Exploring the limits of transfer learning with a unified text-to-text transformer. arXiv preprint arXiv:1910.10683 (2019). · arXiv
- Raphael Tang , Yao Lu , Linqing Liu , Lili Mou , Olga Vechtomova , and Jimmy Lin . 2019. Distilling task-specific knowledge from BERT into simple neural networks. arXiv preprint arXiv:1903.12136 ( 2019 ). Raphael Tang, Yao Lu, Linqing Liu, Lili Mou, Olga Vechtomova, and Jimmy Lin. 2019. Distilling task-specific knowledge from BERT into simple neural networks. arXiv preprint arXiv:1903.12136 (2019). · arXiv
Cited by
Items 501–600 of 1,023
Improving Neural Models for the Retrieval of Relevant Passages to Geographical Queries
Coelho, J., Magalhães, J., & Martins, B. (2021) · ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems (SIGSPATIAL GIS) · doi:10.1145/3474717.3483960 · Cited by 5
Diagnostic Evaluation of Policy-Gradient-Based Ranking
Yu, H.-T., Huang, D., Ren, F., & Li, L. (2021) · Electronics · doi:10.3390/electronics11010037 · Cited by 7
Learning to Enrich Query Representation with Pseudo-Relevance Feedback for Cross-lingual Retrieval
Chandradevan, R., Yang, E., Yarmohammadi, M., & Agichtein, E. (2022) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3477495.3532013 · Cited by 6
“Dense retrieval models, such as ColBERT [10], ANCE [27] and DPR [9], have provided substantial improvements in the neural information retrieval field by enabling documents to be encoded prior to the search, and their counterparts CLIR have also been investigated [17].”
MethodTransformer-Encoder-Based Mathematical Information Retrieval
Reusch, A., Thiele, M., & Lehner, W. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-13643-6_14 · Cited by 5
Dense Re-Ranking with Weak Supervision for RDF Dataset Search
Chen, Q., Huang, Z., Zhang, Z., Luo, W., Lin, T., Shi, Q., & Cheng, G. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-47240-4_2 · Cited by 5
An intention multiple-representation model with expanded information
Hu, J., Peng, J., Zhang, W., Qi, L., Hu, M., & Zhang, H. (2021) · Computer Speech & Language · doi:10.1016/j.csl.2021.101196 · Cited by 5
Li, Y., Fu, X., Verma, G., Buitelaar, P., & Liu, M. (2025) · arXiv · doi:10.48550/arxiv.2510.24476 · Cited by 4
“[56], TF-IDF [120], SPLADE [28] Document retrieval, first-stage recall, resource-limited settings Dense Retrievers Dense vector encoding; similarity via approximate nearest neighbor search DPR[46], Contriever[39], ANCE[158] Open-domain QA, semantic similarity search, cross-lingual retrieval HybridRetrievers Combine sp…”
VStream: A Distributed Streaming Vector Search System
Gong, S., Sun, H., Fang, Z., Liu, L., Chen, L., & Gao, Y. (2025) · Proceedings of the VLDB Endowment · doi:10.14778/3725688.3725692 · Cited by 6
Integrating External Knowledge with LLMs: A Systematic Review of RAG Approaches
Mikulić, I., Vlaić, M., Delač, G., Šilić, M., & Vladimir, K. (2025) · International Convention on Information and Communication Technology, Electronics and Microelectronics · doi:10.1109/mipro65660.2025.11131735 · Cited by 6
“ColBERT [42] employs a BERT-based model to generate token embeddings and identifies relevant documents through a per-token MaxSim operation.”
LLM-Powered Intent-Driven Configuration Generation for Multi-Vendor Networks
Wang, J., He, B., Zhao, J., Xuan, Y., Sun, H., Qi, Q., Liang, J., Zhuang, Z., & Liao, J. (2026) · IEEE Transactions on Network and Service Management · doi:10.1109/tnsm.2026.3675409 · Cited by 10
“• ColBERT: We also consider a late interaction baseline ColBERT [75], which improves the representation-based approach by performing interaction between the intent and syntaxes.”
Approximate Distance-Comparison-Preserving Symmetric Encryption
Fuchsbauer, G., Ghosal, R., Hauke, N., & O’Neill, A. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-14791-3_6 · Cited by 8
Improving Conversational Passage Re-ranking with View Ensemble
Ju, J.-H., Lin, S.-C., Tsai, M.-F., & Wang, C. (2023) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3539618.3592002 · Cited by 4
ELICE: Embedding Language through Informative Contrastive-Encoder
Chiao, C.-S., & Lin, D.-T. (2023) · Expert Systems with Applications · doi:10.1016/j.eswa.2023.120523 · Cited by 4
End-to-End Query Term Weighting
Samel, K., Li, C., Kong, W., Chen, T., Zhang, M., Gupta, S., Khadanga, S., Xu, W., Wang, X., Kolipaka, K., et al. (2023) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3580305.3599815 · Cited by 5
Enhancing Knowledge Sharing Workshops with Natural Language Processing in Maintenance Work
Ogawa, R., Inoue, M., & Uchihira, N. (2024) · International Technical Conference on Circuits/Systems, Computers and Communications (ITC-CSCC) · doi:10.1109/itc-cscc62988.2024.10628182 · Cited by 4
Self-explanatory and Retrieval-augmented LLMs for Financial Sentiment Analysis
Pallucchini, F., Zhang, X., Mao, R., & Cambria, E. (2025) · ACM Symposium on Applied Computing · doi:10.1145/3672608.3707894 · Cited by 6
“For such reason, previous studies have found empirical success with phrase retrieval or late-interaction models, which support more granular and expressive representations of the retrieval corpus [15–17].”
Generalized Contrastive Learning for Multi-Modal Retrieval and Ranking
Zhu, T., Jung, M. C., & Clark, J. (2025) · International World Wide Web Conference (WWW) · doi:10.1145/3701716.3715227 · Cited by 7
“Some datasets offer 3-5 levels of graded relevance [6,36,41], introducing gradations such as directly related, indirectly related, and non-related.”
Towards Comprehensive Legal Document Analysis: A Multi-Round RAG Approach
Zhang, W., Zhou, H., Zhou, Q., Li, Y., Liu, Y., Lou, J., Wu, C., & Li, J. (2025) · International Conference on Multimedia Retrieval · doi:10.1145/3731715.3733451 · Cited by 10
“Further progress was made with introducing the ColBERT model [9], which integrates BERT-based contextual embeddings with efficient late interaction mechanisms.”
QDER: Query-Specific Document and Entity Representations for Multi-Vector Document Re-Ranking
Chatterjee, S., & Dalton, J. (2025) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3726302.3730065 · Cited by 6
R 2 LLMs: Retrieval and Ranking with LLMs
Zuccon, G., Zhuang, S., & Ma, X. (2025) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3726302.3731689 · Cited by 13
Variations in Relevance Judgments and the Shelf Life of Test Collections
Parry, A., Fröbe, M., Scells, H., Schlatt, F., Faggioli, G., Zerhoudi, S., MacAvaney, S., & Yang, E. (2025) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3726302.3730308 · Cited by 15
“In principle, an optimal system would approach a score of 1 for a normalized metric (assuming 𝑘 relevant documents exist).”
From Keywords to Concepts: A Late Interaction Approach to Semantic Product Search on IKEA.com
Gill, A. S., Patel, S., Varga, P., Miller, P., & Athanasiadis, S. (2025) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3726302.3731948 · Cited by 4
Content-Based 3D Image Retrieval and a ColBERT-Inspired Re-ranking for Tumor Flagging and Staging
Jush, F. K., Vogler, S., & Lenga, M. (2025) · Journal of Imaging Informatics in Medicine · doi:10.1007/s10278-025-01598-0 · Cited by 5
Sorour, S. E., & Almusallam, N. (2025) · Alexandria Engineering Journal · doi:10.1016/j.aej.2025.09.070 · Cited by 5
Abdallah, A., Abdalla, M. I., Piryani, B., Mozafari, J., Ali, M., & Jatowt, A. (2025) · ACM International Conference on Information and Knowledge Management (CIKM) · doi:10.1145/3746252.3761484 · Cited by 10
Where Does Legal AI Fail? Evaluating RAG Pipelines
Kim, Y., & Lee, W. (2025) · ACM International Conference on Information and Knowledge Management (CIKM) · doi:10.1145/3746252.3761151 · Cited by 5
On the Use of LLMs for Relevance Labelling
Alaofi, M., Thomas, P., Scholer, F., & Sanderson, M. (2026) · ACM Transactions on Information Systems · doi:10.1145/3788872 · Cited by 8
Unsupervised Dense Retrieval with Conterfactual Contrastive Learning
Chen, H., Ai, Q., Zhou, Y., Wang, X., Liu, Y., Fen, L., & Liu, Q. (2026) · ACM International Conference on Web Search and Data Mining (WSDM) · doi:10.1145/3773966.3778015 · Cited by 5
Ge, Y., Guo, Y., Das, S., & Sarker, A. (2026) · npj Artificial Intelligence · doi:10.1038/s44387-025-00062-2 · Cited by 7
Sivanaiah, R., Angel Deborah, S., Rajendram, S. M., Mirnalinee, T. T., Singh, A. P., Gupta, A., & Nanda, A. (2021) · International Workshop on Semantic Evaluation (SemEval) · doi:10.18653/v1/2021.semeval-1.167 · Cited by 4
“We have used Contextualized Late Interaction over BERT (ColBERT) (Khattab and Zaharia, 2020).”
MethodA Time-Optimized Content Creation Workflow for Remote Teaching
Hofstätter, S., Althammer, S., Sertkan, M., & Hanbury, A. (2022) · ACM Technical Symposium on Computer Science Education (SIGCSE) · doi:10.1145/3478431.3499421 · Cited by 4
“Taking off in 2019 the information retrieval research field began an enormous paradigm shift towards utilizing BERT-based language models [10] in various forms [14, 16, 17, 23] to great effect with huge leaps in quality improvements for search results using large-scale data [3, 8, 9].”
BackgroundPassage Retrieval on Structured Documents Using Graph Attention Networks
Albarede, L., Mulhem, P., Goeuriot, L., Pape, C. L., Marié, S., & Chardin-Segui, T. (2022) · Lecture notes in computer science · doi:10.1007/978-3-030-99739-7_2 · Cited by 8
On the Effects of Regional Spelling Conventions in Retrieval Models
Chari, A., MacAvaney, S., & Ounis, I. (2023) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3539618.3592030 · Cited by 14
BLADE: Combining Vocabulary Pruning and Intermediate Pretraining for Scaleable Neural CLIR
Nair, S., Yang, E., Lawrie, D., Mayfield, J., & Oard, D. W. (2023) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3539618.3591644 · Cited by 19
“In the second approach, query and document processing is separated, so only the query representation needs to be computed at query time [30, 31].”
BackgroundCooperative Game Modeling With Weighted Token-Level Alignment for Audio-Text Retrieval
Xin, Y., Wang, B., & Shang, L. (2023) · IEEE Signal Processing Letters · doi:10.1109/lsp.2023.3313090 · Cited by 8
Enhanced Word-Unit Broad Learning System With Sememes
Jiang, Y., Lu, J., Li, T., & Lv, W. (2023) · IEEE Access · doi:10.1109/access.2023.3322439 · Cited by 3
“stream tasks, including information retrieval [1], sentiment analysis [2], and spam detection [3].”
Learning Sparse Lexical Representations Over Specified Vocabularies for Retrieval
Dudek, J. M., Kong, W., Li, C., Zhang, M., & Bendersky, M. (2023) · ACM International Conference on Information and Knowledge Management (CIKM) · doi:10.1145/3583780.3615207 · Cited by 11
“The central challenge is that, unlike later stages (e.g., ranking), the document pool is too large for expensive comparisons between the query and every document at serving time [13] (e.g., with cross-attention).”
BackgroundA Self-Learning Resource-Efficient Re-Ranking Method for Clinical Trials Search
Rybinski, M., Nguyen, V., & Karimi, S. (2023) · ACM International Conference on Information and Knowledge Management (CIKM) · doi:10.1145/3583780.3615174 · Cited by 4
LEAD: Liberal Feature-based Distillation for Dense Retrieval
Sun, H., Liu, X., Gong, Y., Dong, A., Lu, J., Zhang, Y., Yang, L., Majumder, R., & Duan, N. (2024) · ACM International Conference on Web Search and Data Mining (WSDM) · doi:10.1145/3616855.3635774 · Cited by 5
Beyond Topicality: Including Multidimensional Relevance in Cross-encoder Re-ranking
Upadhyay, R., Askari, A., Pasi, G., & Viviani, M. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-56027-9_16 · Cited by 4
Chang, S., Ahn, G.-J., & Park, S. (2024) · IEEE Access · doi:10.1109/access.2024.3382190 · Cited by 6
Knowledge-augmented Methods for Natural Language Generation
Jiang, M., Lin, B. Y., Wang, S., Yi‐chong, X., Yu, W., & Zhu, C. (2024) · SpringerBriefs in computer science · doi:10.1007/978-981-97-0747-8_4 · Cited by 7
Graded Relevance Scoring of Written Essays with Dense Retrieval
Albatarni, S., Eltanbouly, S., & Elsayed, T. (2024) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3626772.3657744 · Cited by 9
Behmanesh, S., Talebpour, A., Shamsfard, M., & Jafari, M. M. M. (2024) · IEEE Access · doi:10.1109/access.2024.3490452 · Cited by 4
“Using ColBERT [62] instead of BERT enables achieving a higher level of accuracy in understanding and retrieving relevant passages.”
Trippas, J. R. (2024) · ACM eBooks · doi:10.1145/3674127.3674135 · Cited by 6
da Costa, L. Y., & Filho, J. B. D. O. E. S. (2024) · Simpósio Brasileiro de Tecnologia da Informação e da Linguagem Humana (STIL) · doi:10.5753/stil.2024.245443 · Cited by 6
“Three retriever models were evaluated: BM25 [Robertson and Zaragoza, 2009], Dense Passage Retriever (DPR) [Karpukhin et al., 2020], and ColBERT [Khattab and Zaharia, 2020].”
Contention aware DRAM caching for CXL-enabled pooled memory
Tirumalasetty, C., & Reddy, N. (2024) · International Symposium on Memory Systems (MEMSYS) · doi:10.1145/3695794.3695808 · Cited by 6
“ML techniques like large language models(LLM) [22, 39, 69, 74, 75], video processing [32, 37, 76] and others, rely on large amounts of data for training and some-times for retrieval.”
Pre-training, Fine-tuning and Re-ranking: A Three-Stage Framework for Legal Question Answering
Ni, S., Cheng, H., & Yang, M. (2025) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49660.2025.10890476 · Cited by 4
“To verify the effectiveness of our proposed method, we compare our PFR-LQA method with several state-of-the-art baselines which are described below: BERT [11], LawFormer [18], DPR [19], ColBERT [20], SimCSE [21].”
LLM-Based Query Expansion with Gaussian Kernel Semantic Enhancement for Dense Retrieval
Pan, M., Xiong, W., Zhou, S., Gao, M., & Chen, J. (2025) · Electronics · doi:10.3390/electronics14091744 · Cited by 5
Few-shot LLM Synthetic Data with Distribution Matching
Ren, J., Du, Z., Wen, Z., Jia, Q., Dai, S., Wu, C., & Dong, Z. (2025) · International World Wide Web Conference (WWW) · doi:10.1145/3701716.3715245 · Cited by 19
Multimedia Graph Codes for Fast and Semantic Retrieval-Augmented Generation
Wagenpfeil, S. (2025) · Electronics · doi:10.3390/electronics14122472 · Cited by 4
A Summary of the ALQAC 2024 Competition
Do, D.-T., Luu, S. T., Pham, T., Võ, T., Chu, N.-H., Chu, Q.-H., Nguyen, C., Nguyen, L.-M., Trieu, A., Nguyen, D. M.-T., et al. (2024) · International Conference on Knowledge and Systems Engineering/Proceedings · doi:10.1109/kse63888.2024.11063484 · Cited by 6
Zhang, L., Sanyal, S., & King, T. H. (2025) · IEEE ... International Conference on Data Mining workshops · doi:10.1109/icdmw69685.2025.00133 · Cited by 4
Interaction-Based Document Matching for Implicit Search Result Diversification
Qin, X., Dou, Z., Zhu, Y., & Wen, J.-R. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-88189-4_1 · Cited by 4
Parameter-Efficient Sparse Retrievers and Rerankers Using Adapters
Pal, V., Lassance, C., Déjean, H., & Clinchant, S. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-28238-6_2 · Cited by 5
“Their studies show that while cross-encoders with LoRA and LoRA+(50% more parameters compared to LoRA) outperform fine-tuning with TwinBERT [18] and ColBERT [13], parameter-efficient meth-ods do not outperform fine-tuning for bi-encoders across all datasets.”
SS-BERT: A Semantic Information Selecting Approach for Open-Domain Question Answering
Fu, X., Du, J., Zheng, H.-T., Li, J., Hou, C., Zhou, Q., & Kim, H. (2023) · Electronics · doi:10.3390/electronics12071692 · Cited by 4
Pappas, D., & Androutsopoulos, I. (2021) · doi:10.18653/v1/2021.acl-long.301 · Cited by 4
“than re-ranking documents retrieved by conventional IR, have also been proposed (Fan et al., 2018; Ai et al., 2018; Khattab and Zaharia, 2020), but none addresses both document and snippet retrieval.”
BackgroundServerless BM25 Search and BERT Reranking.
Anand, M., Zhang, J., Ding, S., Xin, J., & Lin, J. (2021) · Cited by 4
Improving Dense FAQ Retrieval with Synthetic Training
Liu, L., Wu, Q., & Chen, G. (2021) · IEEE International Conference on Network Infrastructure and Digital Content (IC-NIDC) · doi:10.1109/ic-nidc54101.2021.9660603 · Cited by 4
Li, M., & Gaussier, É. (2022) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3477495.3531856 · Cited by 4
Isotropic Representation Can Improve Dense Retrieval
Jung, E., Park, J., Choi, J., Kim, S., & Rhee, W. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-33380-4_10 · Cited by 6
Proxy-based Zero-Shot Entity Linking by Effective Candidate Retrieval
Wiatrak, M., Arvaniti, E., Brayne, A., Vetterle, J., & Sim, A. (2022) · International Workshop on Health Text Mining and Information Analysis (LOUHI) · doi:10.18653/v1/2022.louhi-1.11 · Cited by 4
Althammer, S., Zuccon, G., Hofstätter, S., Verberne, S., & Hanbury, A. (2023) · Annual International ACM SIGIR Conference on Research and Development in Information Retrieval in the Asia Pacific Region (SIGIR-AP) · doi:10.1145/3624918.3625333 · Cited by 9
“The ColBERT [33] method delays the interaction between the query and passage to after the encoding by computing the relevance score as the sum of the maximum similarity scores between all token representations of the query and passage: As for single representation bi-encoder methods, also in…”
BackgroundMethodResultInfluentialSynthetic Target Domain Supervision for Open Retrieval QA
Reddy, R. G., Iyer, B., Sultan, M. A., Zhang, R., Sil, A., Castelli, V., Florian, R., & Roukos, S. (2021) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3404835.3463085 · Cited by 9
“While other competitive methods such as ColBERT [22] exist, DPR offers a number of advantages in real-time settings as well as in our target scenario where retrieval is only a component in a larger ORQA pipeline.”
MethodUHD-BERT: Bucketed Ultra-High Dimensional Sparse Representations for Full Ranking.Preprint
Jang, K.-R., Kang, J., Hong, G., Myaeng, S.-H., Park, J., Yoon, T., & Seo, H.-C. (2021) · arXiv (Cornell University) · Cited by 4
Sequential Attention Module for Natural Language ProcessingPreprint
Zhou, M., Ma, J., Yang, H., Jiang, L., & Mo, Y. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2109.03009 · Cited by 5
“Interaction between tokens has been delayed to reduce the computation in the self-attention mechanism (Humeau et al., 2020; Khattab and Zaharia, 2020).”
Automatic question answering for multiple stakeholders, the epidemic question answering dataset
Goodwin, T. R., Demner‐Fushman, D., Lo, K., Wang, L. L., Dang, H. T., & Soboroff, I. M. (2022) · Scientific Data · doi:10.1038/s41597-022-01533-w · Cited by 5
MRAG: a modular retrieval framework for time-sensitive question answering
Zhang, S., Xue, Y., Zhang, Y., Wu, X., Luu, A. T., & Zhao, C. (2025) · Findings of the Association for Computational Linguistics: EMNLP 2025 · doi:10.18653/v1/2025.findings-emnlp.167 · Cited by 3
A Test Collection for Dataset Retrieval
Kolyada, N., Potthast, M., & Stein, B. (2025) · Lecture notes in computer science · doi:10.1007/978-3-031-88714-7_36 · Cited by 5
Explainable LLM-driven Multi-dimensional Distillation for E-Commerce Relevance Learning
Zhao, G., Zhang, X., Lu, C., Zhao, H., Wu, T., Wang, P., Xu, J., & Zheng, B. (2025) · International World Wide Web Conference (WWW) · doi:10.1145/3701716.3715222 · Cited by 7
“3) ColBERT [20], a representation-based method that further improves the query-item interaction by employing a token-level late interaction mechanism.”
InfluentialYao, S., Hu, J., Zhang, Z., & Liu, D. (2025) · Neural Networks · doi:10.1016/j.neunet.2025.107663 · Cited by 3
Improved IR-Based Bug Localization with Intelligent Relevance Feedback
Samir, A. M., & Rahman, M. M. (2025) · IEEE International Conference on Program Comprehension (ICPC) · doi:10.1109/icpc66645.2025.00065 · Cited by 13
“A recent technique, FBL-BERT [86] uses a BERT-based model, ColBERT [87], for document scoring with late interaction.”
Reproducibility, Replicability, and Insights into Visual Document Retrieval with Late Interaction
Qiao, J., Ju, J.-H., Ma, X., Kanoulas, E., & Yates, A. (2025) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3726302.3730285 · Cited by 3
“Unlike the single-vector approach used in DSE, ColPali [10], a novel approach for visual document retrieval, integrates the late-interaction mechanism [13] into relevance score estimation.”
InfluentialMovin, M., & Hauff, C. (2025) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3726302.3730119 · Cited by 4
“…the focus was on the fine-tuning of the highly effective (but high-latency) cross-encoder models [8, 35, 38], or the less performant but more scalable bi-encoder models [18, 22, 24, 53] for ad-hoc reranking as well as the generation of training data [2, 9, 20] and knowledge distillation [40, 49].”
Generate-Distill: Training Cross-Language IR Models with Synthetically-Generated Data
Lawrie, D., Kayi, E. S., Yang, E., Mayfield, J., Oard, D. W., & Miller, S. (2025) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3726302.3730201 · Cited by 4
“UMR generally underperforms DPR-MT, which prior work [15, 19] has shown to be, in turn, significantly worse than Translate-Distill [37].”
CoDIME: A Counterfactual Approach for Dimension Importance Estimation through Click Logs
Faggioli, G., Ferro, N., Perego, R., & Tonellotto, N. (2025) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3726302.3729926 · Cited by 6
Math Formula Graph Retrieval Using Contrastive Learning Over Visual and Semantic Embeddings
Amador, B., & Zanibbi, R. (2025) · ACM SIGIR International Conference on Theory of Information Retrieval (ICTIR) · doi:10.1145/3731120.3744589 · Cited by 6
Projection-Displacement-Based Query Performance Prediction for Embedded Space of Dense Retrievers
Datta, S., Faggioli, G., Ferro, N., Ganguly, D., Muntean, C. I., Perego, R., & Tonellotto, N. (2025) · ACM Transactions on Information Systems · doi:10.1145/3765617 · Cited by 3
“To demonstrate the efficacy of their approach, they utilize WRIG to predict performance in BM25, four variations of DRMM [32], and the initial stage neural IR model, ColBERT [39].”
Adaptive Hybrid Retrieval-Augmented Generation with Context-Aware Confidence Control
Sun, Y. (2025) · International Conference on Electronic Information Technology (EIT) · doi:10.1109/eit67313.2025.11231898 · Cited by 3
“ColBERT [8] Late Interaction retrieval model, using the officially released NQ weights.”
Chifu, A.-G., Dejean, S., Garouani, M., Mothe, J., Ortiz, D., & Ullah, M. Z. (2026) · ACM Transactions on Information Systems · doi:10.1145/3774427 · Cited by 9
“These contrasting paradigms introduce complexities for QPP, as state-of-the-art (SOTA) predictors (e.g., NQC [9]) and LETOR-based features were primarily designed for sparse systems and may fail to capture the nuances of dense retrieval.”
VLMAR: Maritime scene anomaly detection via retrieval-augmented vision-language models
Wang, S., Yang, C., & Cai, C. (2025) · Journal of Visual Communication and Image Representation · doi:10.1016/j.jvcir.2025.104669 · Cited by 5
Hu, C., Huang, Y., Kuang, J., Dai, B., Peng, Y., Xiao, Y., & Su, Y. (2026) · Electronics · doi:10.3390/electronics15030541 · Cited by 3
Bayesian RAG: uncertainty-aware retrieval for reliable financial question answering
Ngartera, L., Nadarajah, S., & Rodoumta, K. (2026) · Frontiers in Artificial Intelligence · doi:10.3389/frai.2025.1668172 · Cited by 6
Hybrid retrieval generation for structured reasoning with large language models
Muthusami, R., & Saritha, K. (2026) · Discover Artificial Intelligence · doi:10.1007/s44163-026-01059-9 · Cited by 4
Interpreting Patient Descriptions using Distantly Supervised Similar Case Retrieval
Alghanmi, I., Espinosa-Anke, L., & Schockaert, S. (2022) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3477495.3532003 · Cited by 5
“[32] start from a pre-trained ColBERT model [33] to get an”
MethodKnowledge-grounded Dialog State Tracking
Yu, D., Wang, M., Cao, Y., Shafey, L. E., Izhak Shafran, J., & Soltau, H. (2022) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2022.findings-emnlp.250 · Cited by 5
Reproducibility Analysis and Enhancements for Multi-aspect Dense Retriever with Aspect Learning
Bi, K., Sun, X., Guo, J., & Cheng, X. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-56066-8_17 · Cited by 7
Shi, H., & Sakai, T. (2024) · IEEE Access · doi:10.1109/access.2024.3378518 · Cited by 3
“While some neural IR models, such as ColBERT [28], shift towards soft semantic matching over all query-document terms and achieve commendable performance in Ad-hoc retrieval, they compromise on the computational efficiency of exact match systems.”
MethodInDi: Informative and Diverse Sampling for Dense Retrieval
Cohen, N., Indelman, H. C., Fairstein, Y., & Kushilevitz, G. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-56063-7_16 · Cited by 3
DISKCO : Disentangling Knowledge from Cross-Encoder to Bi-Encoder
Ankith, M. S., Bhattacharya, A., Gandhi, A., Huddar, V., Saroop, A., & Bhagat, R. (2024) · International World Wide Web Conference (WWW) · doi:10.1145/3589335.3648333 · Cited by 4
“It has an architecture similar to Bi-Encoder, but instead of using a simple distance metric like cosine similarity, it utilizes late interaction layers to combine the embeddings returned by each Bi-encoder arm, and compute the final similarity scores.”
What Is Wrong with My Model? Identifying Systematic Problems with Semantic Data Slicing
Yang, C., Hong, Y., Lewis, G. A., Wu, T., & Kästner, C. (2024) · IEEE/ACM International Conference on Automated Software Engineering (ASE) · doi:10.1145/3691620.3695033 · Cited by 5
Causal Probing for Dual Encoders
Wallat, J., Hinrichs, H., & Anand, A. (2024) · ACM International Conference on Information and Knowledge Management (CIKM) · doi:10.1145/3627673.3679556 · Cited by 4
Bac, N. X., Sang, L. V., Nguyen, V. D., Le, L. Q., & Thinh, D. D. (2024) · Journal of Military Science and Technology · doi:10.54939/1859-1043.j.mst.99.2024.109-118 · Cited by 3
Verberne, S. (2024) · ACM eBooks · doi:10.1145/3674127.3674141 · Cited by 5
Tonellotto, N. (2024) · ACM eBooks · doi:10.1145/3674127.3674130 · Cited by 5
Fair Exposure Allocation Using Generative Query Expansion
Jaenich, T., McDonald, G., & Ounis, I. (2025) · Lecture notes in computer science · doi:10.1007/978-3-031-88717-8_20 · Cited by 5
“– ColBERT [16]: A neural ranking model that uses contextualised late interaction for efficient and effective document retrieval.”
ColBERT-Serve: Efficient Multi-stage Memory-Mapped Scoring
Huang, K., Venkatesh, T., Dingankar, U., Mallia, A., Campos, D., Jiao, J., Potts, C., Zaharia, M., Boahen, K. A., Khattab, O., et al. (2025) · Lecture notes in computer science · doi:10.1007/978-3-031-88717-8_3 · Cited by 3
“For MS MARCO, we report development results on Dev, on which we tune α for all datasets, and report evaluation results on the held-out evaluation set used by the ColBERT authors [7,17].”
Self-citationImproving Vietnamese Legal Document Retrieval Using Synthetic Data
Son, P. T., Hieu, N. D., An, N. D., & Sang, D. V. (2025) · Communications in computer and information science · doi:10.1007/978-981-96-4291-5_31 · Cited by 4
Stack Trace Deduplication: Faster, More Accurately, and in More Realistic Scenarios
Shibaev, E., Sushentsev, D., Golubev, Y., & Khvorov, A. (2025) · IEEE International Conference on Software Analysis, Evolution and Reengineering (SANER) · doi:10.1109/saner64311.2025.00054 · Cited by 4
“3) Creating embeddings of stack traces with cross-encoder: A cross-encoder is a powerful mechanism in machine learning, specifically designed to process pairs of inputs simultaneously to produce a single output, which is highly beneficial for the retrieval task [18], [34]–[36].”
InteractRank: Personalized Web-Scale Search Pre-Ranking with Cross Interaction Features
Khandagale, S., Juneja, B., Agarwal, P., Subramanian, A., Yang, J., & Wang, Y. (2025) · International World Wide Web Conference (WWW) · doi:10.1145/3701716.3715239 · Cited by 3
“While it learns a dense vector for query and item using InfoNCE loss [12], it promotes early interaction through a MaxSim [19] based score function on the components of the learned vectors and hidden layer representations.”
Metadata
- Authors
- Khattab, O., & Zaharia, M.
- Affiliations
- Stanford University (US)
- Venue
- Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval
- Publisher
- ACM
- Published
- 2020-07-25
- Pages
- 39-48
- OpenAlex
- W3021397474
- Cited by
- 1,301
- References
- 21
- Citations by year
- 2026: 346 · 2025: 277 · 2024: 197 · 2023: 184 · 2022: 143 · 2021: 135 · 2020: 18
- FWCI
- 90.73
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) (publishedVersion)
- Funders
- NSF (CNS-1651570)
- Topics
- Physical Sciences › Computer Science › Information Systems › Information Retrieval and Search Behavior
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Image and Video Retrieval Techniques
- Keywords
- Colbert, BERT, document ranking, information retrieval, query-document similarity, efficient retrieval
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.