Pith. sign in

REVIEW 3 major objections 5 minor 47 references

Neighborhood-Aware Dual Biomedical Entity Linking

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A three-stage framework grounds biomedical mentions to ontologies, besting prior methods on four of five benchmarks.

desk verdict PILOT is a well-engineered BioEL system with honest ablations, but the headline state-of-the-art claim currently rests on baseline numbers the authors copied rather than re-ran, and the AAP statistics are internally inconsistent. read the letter →

arxiv 2608.04144 v1 pith:DTIVBDWM submitted 2026-08-04 cs.IR cs.CLcs.LG

Yicheng Tao , Jie Liu This is my paper · ORCID
classification cs.IRcs.CLcs.LG
keywords biomedicalentitylinkingknowledgebasegroundingontology-awareretrievalqueryreformulationrerankingscorefusionlargelanguagemodelsUMLS
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper claims that a retrieve-and-rerank framework called PILOT solves biomedical entity linking better than existing discriminative, generative, and LLM-based methods, with the best average accuracy across five standard benchmarks and an order-of-magnitude speedup over the strongest LLM baseline. The design injects ontological structure on both sides of retrieval: LLM-generated standard names and hypernyms on the query side, and parent/child embedding pooling on the knowledge-base side. It then reranks candidates from two complementary views, surface form and context, and fuses the two scores. If correct, practical and efficient grounding of clinical and scientific mentions to curated ontologies becomes achievable. The claimed gains are consistent across datasets rather than driven by any single benchmark.

What carries the argument

The framework has three coupled stages. The neighborhood-aware retriever computes a query embedding as a convex combination of the mention embedding, an LLM-generated standardized name, and a hypernym, and computes each KB entity embedding with a one-hop Rocchio update that averages in its parents and children, so retrieval pulls in ontologically adjacent entities. The dual reranker scores the top-K candidates from a surface-form view—a precision-configured second pass of the same encoder—and a contextual view, a pointwise generative cross-encoder that reads the mention in its sentence plus a textual rendering of the candidate's synonyms and one-hop ontology neighbors. The two score vectors are z-normalized and fused as $(1-\alpha)z(s_{sf}) + \alpha z(s_{ctx})$, with $\alpha$ chosen on a development set. The pointwise design keeps per-candidate input length constant, making deep pools affordable.

What would settle it

Re-run the published baselines (ANGEL, LLM4BioEL, GenBioEL, Prompt-BioEL) on the same five datasets under the exact data splits, KB subsets, and R@1 computation used for PILOT, using the same 10-fold cross-validation for AAP, and check whether the average deltas of +2.1 points persist.

Watch

Extended reading notes

Core claim

PILOT achieves the highest average Recall@1 across NCBI, BC5CDR, COMETA, AAP, and MM-ST21pv, beating the strongest prior method ANGEL by 2.1 points (88.8 vs. 86.7) and the leading LLM-based baseline LLM4BioEL by 2.1 points on the first three datasets. It is best on four of the five benchmarks, with NCBI the only exception, where it is a close second. At inference it is 7-11x faster than LLM4BioEL, reaching 1.8-5.2 mentions per GPU-second despite reranking a deep candidate pool of 50. Ablations show all three components—query reformulation, ontology-pooled embeddings, and the dual reranker with score fusion—contribute, with the contextual reranker being the largest single contributor overall and the surface-form reranker mattering most on NCBI.

Load-bearing premise

The claimed state of the art rests on assuming the baseline numbers taken from the ANGEL and LLM4BioEL papers were produced under the identical evaluation protocol—same data splits, KB subsets, and metric computation—as PILOT's runs, and the paper does not verify this by re-running those baselines.

Editorial extensions

If this is right

  • Biomedical entity linking can be solved by a decoupled retrieve-and-rerank pipeline rather than end-to-end generative decoding, and the rerank stage is where most accuracy is won.
  • Deep candidate pools (K=50) become feasible with pointwise reranking, and depth is exactly what improves accuracy on the hardest datasets like COMETA and MM-ST21pv.
  • Injecting ontological structure on both query and KB sides—rather than only one—is what lifts retrieval recall for mentions with irregular surface forms.
  • The 7-11x throughput advantage over LLM4BioEL makes the approach practical for time-sensitive clinical and literature-scale use.
  • Training the reranker on a mix of surface-form semi-hard negatives and its own top errors is necessary to suppress confusable lexical neighbors.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same neighborhood-aware recipe would likely transfer to other KB-grounded tasks such as chemical or gene normalization, and to ontologies that lack a dense is-a hierarchy only partially.
  • Because all weights and the fusion weight are tuned per dataset on a labeled development set, the reported numbers probably do not reflect zero-shot or cross-corpus performance; a testable extension is to report transfer without in-domain tuning.
  • The pointwise reranker trades away direct candidate comparison; adding a lightweight pairwise or setwise second pass within the same fusion scheme might close the remaining gap on near-identical surface forms.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper presents PILOT, a three-stage biomedical entity linking framework: a neighborhood-aware retriever that combines generative query reformulation (standardized name and hypernym) with one-hop Rocchio pooling over ontology neighbors; a dual reranker consisting of a surface-form second pass and a pointwise generative cross-encoder; and z-normalized score fusion. It reports test R@1 on five BioEL benchmarks, claiming the best average result and a +2.1 point improvement over ANGEL, together with a 7-11x throughput advantage over LLM4BioEL. The paper also provides ablations, hyperparameter sensitivity analyses, and a limitations section. The central validation is empirical; no proofs, code, or evaluation harness are provided.

Significance. If the empirical claims hold, PILOT is a practically valuable contribution: the pointwise reranker design makes deep candidate pools affordable, and the ablations plus sensitivity analyses give a clear picture of which components matter and where. The limitations section is honest and the framing of the two reranker views is well motivated. However, the state-of-the-art claim rests entirely on baseline comparability and protocol consistency that are not currently demonstrated. The paper does not ship machine-checked proofs or reproducible code, and the AAP evaluation is internally inconsistent, so the headline +2.1 average gain is not yet established.

major comments (3)
  1. [§4.2, Table 1] The headline result — best average R@1 and +2.1 over ANGEL — is computed from baseline numbers taken verbatim from the ANGEL and LLM4BioEL papers, and the paper does not re-run any baseline under PILOT's protocol. The statement that PILOT adopts the same evaluation protocol as ANGEL is not evidence that the copied scores were produced with identical data splits, KB subsets, candidate pools, or metric computations; for instance, §4.6 notes that LLM4BioEL uses retrieval depth K in [1,10] and Prompt-BioEL uses K=6, while PILOT uses K=50. Without a reproduced evaluation harness or protocol documentation, the +2.1 average gain is an assumption rather than a measured result.
  2. [§4.1, Table 4, Appendix B] The AAP evaluation protocol is internally inconsistent. Section 4.1 says that because AAP provides no official test split, PILOT uses 10-fold cross-validation with per-fold hyperparameter selection, and Appendix B confirms that query-mix, pooling, and fusion weights are selected per fold. Yet Table 4 reports a single fixed Train/Dev/Test split of 15,665/793/866, while the Introduction describes AAP as containing 8,662 social media phrases (15,665+793+866 = 17,324). The paper must specify the actual fold sizes, the relation between phrases and mention instances, and how per-fold hyperparameter selection is consistent with comparing against ANGEL's AAP score, which may have been produced under a different split or a single development set. As written, the AAP R@1 of 91.7 and the +1.5 gap over ANGEL are not interpretable.
  3. [§4.4, Tables 1 and 3, §4.6] No error bars, confidence intervals, or significance tests are reported for any test R@1. This is particularly consequential for AAP, where 10-fold cross-validation yields a natural variance estimate, and for NCBI's 960-sample test set, where a 0.4-point gap is plausibly within noise. The absence of variance reporting makes it impossible to assess whether the claimed 2.1-point average improvement is statistically reliable, and the per-fold hyperparameter tuning on AAP compounds the issue. I would like to see at least standard deviations for the 10-fold AAP result and a statement about the significance of the headline gaps.
minor comments (5)
  1. [Table 1] The LLM4BioEL row renders '93.892.4' without a separator, making the two baseline scores difficult to read; please add a space or a delimiter.
  2. [§4.3] It is not stated whether Qwen3-4B-Instruct is used zero-shot for query reformulation or whether it is fine-tuned; please clarify the status of the 'instruction-tuned LLM' used in the retrieval stage.
  3. [§4.1 and Table 4] The fixed splits listed in Table 4 and the per-fold AAP procedure described in Appendix B should be reconciled in one place, rather than requiring the reader to cross-check three sections to understand the AAP evaluation.
  4. [Figure 5] The deltas printed next to the retrieval-depth sweep points are not defined in the caption; please state explicitly what they represent (e.g., change in dev R@1 relative to K=50).
  5. [Reproducibility] The paper does not provide code or an evaluation harness; releasing the prompts, the exact candidate-pool construction, and the benchmark-specific evaluation scripts would substantially strengthen the empirical claims.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity; the sole self-citation is illustrative and not load-bearing.

full rationale

PILOT's central claims are empirical and measured against held-out test labels of five external BioEL benchmarks. The retriever's query mix (Eq. 1), ontology-pooled KB embeddings (Eq. 2), and score fusion (Eq. 3) are defined directly from the mention, context, and KB structure; none of these quantities is defined in terms of the reported R@1 numbers, and the hyperparameters (w, lambda, alpha) are selected on development sets rather than fitted to the test predictions. The ablations and sensitivity analyses provide independent component-level support. The only self-citation is in Section 1, where 'patient-record normalization (Tao et al., 2026)' is cited as an example downstream application; it is not used to justify any method component, baseline, or uniqueness claim, so it is not load-bearing. The main caveat is Table 1's baseline comparability: the paper states that 'the results of the discriminative and generative methods are directly taken from the ANGEL paper, and those of the LLM-based methods are from the LLM4BioEL paper,' so the claimed +2.1 average improvement depends on those baselines having used identical evaluation protocols (splits, KB subsets, candidate pools, metric computation). That is a verification and reproducibility concern, not a circularity, because PILOT's own numbers are measured rather than derived from the baselines. No equation or claim reduces by construction to its inputs, and no self-citation chain forces the framework's choices. The limitations section also explicitly acknowledges the supervised, non-zero-shot nature of the evaluation, which further confirms that the results are empirical rather than definitional.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

PILOT introduces no new theoretical entities. The main burden of the empirical claim is carried by per-dataset hyperparameter tuning (query-mix weights, pooling weight, fusion weight) and by the unverified assumption that copied baseline numbers are comparable. The reliance on in-domain supervision is explicitly acknowledged in the limitations.

free parameters (6)
  • query-mix weights (wm, ws, wh) = Per-dataset, e.g., NCBI retriever: 0.8/0.2/0.0, reranker: 0.9/0.0/0.1; see Figures 2-3 and Appendix B
    Weights for mixing mention, standardized-name, and hypernym embeddings in the retriever and surface-form reranker; tuned on the development set to maximize R@K/R@1.
  • pooling weight lambda = Per-dataset, range [0.1, 0.5]
    Weight for the one-hop Rocchio pooling over parent and child entity embeddings; tuned on the development set.
  • fusion weight alpha = Per-dataset, e.g., NCBI 0.65, AAP average 0.68, overall range [0.55, 0.90]
    Convex combination weight for fusing surface-form and contextual reranker scores; tuned on the development set by R@1.
  • retrieval depth K = 50
    Candidate pool size; fixed after sensitivity sweep showing saturation at K=50.
  • hard negative count and band margins = n=8, delta_lo=0.30, delta_up=0.02, n_sh=5
    Number of negatives per positive and the semi-hard band margins used to mine hard negatives for the contextual reranker; chosen by hand.
  • LoRA and training hyperparameters = rank 16, scaling factor 32, E=2, lr=5e-5, batch size 8
    Fine-tuning configuration for the contextual reranker; fixed across datasets.
assumptions (5)
  • domain assumption The target KB's 'is-a' hierarchy is sufficiently complete and accurate for one-hop pooling to improve retrieval.
    Section 3.1 uses parent and child embeddings in Eq. (2). The limitations section explicitly notes the hierarchy may be incomplete or unavailable in some KBs.
  • domain assumption The instruction-tuned LLM produces accurate standardized names and hypernyms for arbitrary mentions.
    Section 3.1 relies on these reformulations in Eq. (1). No automatic quality check or human evaluation of the reformulations is reported.
  • domain assumption Baseline results copied from the ANGEL and LLM4BioEL papers were produced under the same evaluation protocol as PILOT (data splits, KB subsets, metric computation).
    Section 4.2 states baseline numbers are directly taken from prior papers without re-running. Any protocol mismatch would change the comparison.
  • domain assumption Test sets are representative and R@1 is a sufficient metric for comparing BioEL systems.
    Section 4.1 reports only R@1; no significance tests or additional metrics are used.
  • standard math Standard mathematical background (cosine similarity, convex combination, z-normalization) is valid.
    Equations (1)-(3) use elementary linear algebra and statistics without proof.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Neighborhood-Aware Dual Biomedical Entity Linking." pith.science (2026). https://pith.science/paper/DTIVBDWM

@misc{pith2026260804144,
  author       = {Pith},
  title        = {Pith review of: Neighborhood-Aware Dual Biomedical Entity Linking},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DTIVBDWM}},
  note         = {Machine review of arXiv:2608.04144}
}
read the original abstract

Biomedical entity linking grounds mentions in clinical and scientific text to entities in a curated knowledge base (KB) with ontological structure, which supports downstream applications such as literature-scale information extraction and patient-record normalization. The task has several challenges at once: the KB contains large numbers of entities, mentions are often ambiguous, and gold labels follow annotation conventions specific to each corpus. To address these challenges, we propose PILOT, a three-stage framework made up of neighborhood-aware retrieval, dual reranking, and score fusion. The retriever injects ontological structure from both the query and KB side, by reformulating mentions and pooling entity embeddings. The retrieved pool is then scored from two complementary views, one over surface forms and one over context, and fused together. PILOT achieves the state of the art on average across five widely-used benchmarks and remains efficient at inference.

Figures

Figures reproduced from arXiv: 2608.04144 by the authors.

Figure 1
Figure 1. Overview of PILOT. The two enrichments share the same principle— injecting the ontological neighborhood, one on the query side and one on the KB side—but are mechanically distinct (generative reformulation vs. embedding pooling). Together, they improve the recall of the retriever, with the query-mix weights (wm, ws, wh) and pooling weight λ tuned on the development set to maximize Recall@K (R@K). 3.2 Dual reranking … view at source ↗
Figure 2
Figure 2. PILOT’s sensitivity to the query-mix weights [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. PILOT’s sensitivity to the pooling weight [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: PILOT’s sensitivity to the fusion weight [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]
Figure 5
Figure 5. Figure 5: PILOT’s sensitivity to the retrieval depth [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

47 extracted references · 33 canonical work pages

  1. [1]

    Proceedings of the 58th annual meeting of the association for computational linguistics , pages=

    Biomedical entity representations with synonym marginalization , author=. Proceedings of the 58th annual meeting of the association for computational linguistics , pages=

    work page

  2. [2]

    Proceedings of the 2021 conference of the North American chapter of the association for computational linguistics: human language technologies , pages=

    Self-alignment pretraining for biomedical entity representations , author=. Proceedings of the 2021 conference of the North American chapter of the association for computational linguistics: human language technologies , pages=

    2021

  3. [3]

    Findings of the Association for Computational Linguistics: EMNLP 2021 , pages=

    BERT might be overkill: A tiny but effective biomedical entity linker based on residual convolutional neural networks , author=. Findings of the Association for Computational Linguistics: EMNLP 2021 , pages=

    work page 2021

  4. [4]

    Findings of the Association for Computational Linguistics: EMNLP 2022 , pages=

    Knowledge-rich self-supervision for biomedical entity linking , author=. Findings of the Association for Computational Linguistics: EMNLP 2022 , pages=

    work page 2022

  5. [5]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Improving biomedical entity linking with cross-entity interaction , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    work page

  6. [6]

    Proceedings of the 58th annual meeting of the association for computational linguistics , pages=

    BART: Denoising sequence-to-sequence pre-training for natural language generation, translation, and comprehension , author=. Proceedings of the 58th annual meeting of the association for computational linguistics , pages=

  7. [7]

    Proceedings of the 21st Workshop on Biomedical Language Processing , pages=

    BioBART: Pretraining and evaluation of a biomedical generative language model , author=. Proceedings of the 21st Workshop on Biomedical Language Processing , pages=

    work page

  8. [8]

    Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies , pages=

    Generative biomedical entity linking via knowledge base-guided pre-training and synonyms-aware fine-tuning , author=. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies , pages=

    work page 2022

Show all 47 references
  1. [9]

    Findings of the Association for Computational Linguistics: ACL 2025 , pages=

    Learning from negative samples in biomedical generative entity linking , author=. Findings of the Association for Computational Linguistics: ACL 2025 , pages=

    2025

  2. [10]

    arXiv preprint arXiv:2303.08774 , year=

    Gpt-4 technical report , author=. arXiv preprint arXiv:2303.08774 , year=

    arXiv

  3. [11]

    arXiv preprint arXiv:2501.12948 , year=

    Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning , author=. arXiv preprint arXiv:2501.12948 , year=

    arXiv

  4. [12]

    Proceedings of the Findings of the Association for Computational Linguistics: EMNLP , year=

    Guiding Large Language Models for Biomedical Entity Linking via Restrictive and Contrastive Decoding , author=. Proceedings of the Findings of the Association for Computational Linguistics: EMNLP , year=

  5. [13]

    Journal of biomedical informatics , volume=

    NCBI disease corpus: a resource for disease name recognition and concept normalization , author=. Journal of biomedical informatics , volume=. 2014 , publisher=

    2014

  6. [14]

    Database , volume=

    BioCreative V CDR task corpus: a resource for chemical disease relation extraction , author=. Database , volume=. 2016 , publisher=

    2016

  7. [15]

    Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP) , pages=

    COMETA: A corpus for medical entity linking in the social media , author=. Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP) , pages=

    2020

  8. [16]

    Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (volume 1: long papers) , pages=

    Normalising medical concepts in social media texts by learning semantic representation , author=. Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (volume 1: long papers) , pages=

  9. [17]

    arXiv preprint arXiv:1902.09476 , year=

    Medmentions: A large biomedical corpus annotated with umls concepts , author=. arXiv preprint arXiv:1902.09476 , year=

    1902 arXiv

  10. [18]

    Bulletin of the Medical Library Association , volume=

    Medical subject headings (MeSH) , author=. Bulletin of the Medical Library Association , volume=

  11. [19]

    Nucleic acids research , volume=

    Online Mendelian Inheritance in Man (OMIM), a knowledgebase of human genes and genetic disorders , author=. Nucleic acids research , volume=. 2005 , publisher=

    2005

  12. [20]

    Database , volume=

    MEDIC: a practical disease vocabulary used at the Comparative Toxicogenomics Database , author=. Database , volume=. 2012 , publisher=

    2012

  13. [21]

    Journal of the American Medical Informatics Association , volume=

    The use of SNOMED CT, 2013-2020: a literature review , author=. Journal of the American Medical Informatics Association , volume=. 2021 , publisher=

    2013

  14. [22]

    Nucleic acids research , volume=

    The unified medical language system (UMLS): integrating biomedical terminology , author=. Nucleic acids research , volume=. 2004 , publisher=

    2004

  15. [23]

    The SMART retrieval system: experiments in automatic document processing , year=

    Relevance feedback in information retrieval , author=. The SMART retrieval system: experiments in automatic document processing , year=

  16. [24]

    Findings of the Association for Computational Linguistics: NAACL 2024 , pages=

    Large language models are effective text rankers with pairwise ranking prompting , author=. Findings of the Association for Computational Linguistics: NAACL 2024 , pages=

    2024

  17. [25]

    Transactions of the association for computational linguistics , volume=

    Lost in the middle: How language models use long contexts , author=. Transactions of the association for computational linguistics , volume=

  18. [26]

    Proceedings of the 2023 conference on empirical methods in natural language processing , pages=

    Is ChatGPT good at search? investigating large language models as re-ranking agents , author=. Proceedings of the 2023 conference on empirical methods in natural language processing , pages=

    2023

  19. [27]

    arXiv preprint arXiv:2505.09388 , year=

    Qwen3 technical report , author=. arXiv preprint arXiv:2505.09388 , year=

    arXiv

  20. [28]

    arXiv preprint arXiv:2506.05176 , year=

    Qwen3 embedding: Advancing text embedding and reranking through foundation models , author=. arXiv preprint arXiv:2506.05176 , year=

    arXiv

  21. [29]

    , author=

    Lora: Low-rank adaptation of large language models. , author=. Iclr , volume=

  22. [30]

    Proceedings of the 2024 conference of the North American chapter of the Association for Computational Linguistics: Human language technologies (volume 2: short papers) , pages=

    Beyond yes and no: Improving zero-shot llm rankers via scoring fine-grained relevance labels , author=. Proceedings of the 2024 conference of the North American chapter of the Association for Computational Linguistics: Human language technologies (volume 2: short papers) , pages=

    2024

  23. [31]

    arXiv preprint arXiv:1910.14424 , year=

    Multi-stage document ranking with BERT , author=. arXiv preprint arXiv:1910.14424 , year=

    1910 arXiv

  24. [32]

    Findings of the association for computational linguistics: EMNLP 2020 , pages=

    Document ranking with a pretrained sequence-to-sequence model , author=. Findings of the association for computational linguistics: EMNLP 2020 , pages=

    2020

  25. [33]

    Proceedings of the 46th international ACM SIGIR conference on research and development in information retrieval , pages=

    Rankt5: Fine-tuning t5 for text ranking with ranking losses , author=. Proceedings of the 46th international ACM SIGIR conference on research and development in information retrieval , pages=

  26. [34]

    Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval , pages=

    A setwise approach for effective and highly efficient zero-shot ranking with large language models , author=. Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval , pages=

  27. [35]

    Proceedings of the 62nd annual meeting of the association for computational linguistics (volume 1: long papers) , pages=

    Prp-graph: Pairwise ranking prompting to llms with graph aggregation for effective text re-ranking , author=. Proceedings of the 62nd annual meeting of the association for computational linguistics (volume 1: long papers) , pages=

  28. [36]

    arXiv preprint arXiv:2312.02724 , year=

    Rankzephyr: Effective and robust zero-shot listwise reranking is a breeze! , author=. arXiv preprint arXiv:2312.02724 , year=

    arXiv

  29. [37]

    Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing , pages=

    Conqrr: Conversational query rewriting for retrieval with reinforcement learning , author=. Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing , pages=

    2022

  30. [38]

    Findings of the Association for Computational Linguistics: EMNLP 2023 , pages=

    Large language models know your contextual search intent: A prompting framework for conversational search , author=. Findings of the Association for Computational Linguistics: EMNLP 2023 , pages=

    2023

  31. [39]

    Proceedings of the 2023 conference on empirical methods in natural language processing , pages=

    Query rewriting in retrieval-augmented large language models , author=. Proceedings of the 2023 conference on empirical methods in natural language processing , pages=

    2023

  32. [40]

    Findings of the Association for Computational Linguistics: EMNLP 2024 , pages=

    RaFe: Ranking feedback improves query rewriting for RAG , author=. Findings of the Association for Computational Linguistics: EMNLP 2024 , pages=

    2024

  33. [41]

    Companion Proceedings of the ACM Web Conference 2024 , pages=

    Large language model based long-tail query rewriting in taobao search , author=. Companion Proceedings of the ACM Web Conference 2024 , pages=

    2024

  34. [42]

    Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing , pages=

    Query2doc: Query expansion with large language models , author=. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing , pages=

    2023

  35. [43]

    Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=

    Precise zero-shot dense retrieval without relevance labels , author=. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=

  36. [44]

    Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 2: Short Papers) , pages=

    Corpus-steered query expansion with large language models , author=. Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 2: Short Papers) , pages=

  37. [45]

    Autoregressive Entity Retrieval , booktitle =

    Nicola. Autoregressive Entity Retrieval , booktitle =. 2021 , url =

    2021

  38. [46]

    bioRxiv , pages=

    Building a literature knowledge base towards transparent biomedical AI , author=. bioRxiv , pages=. 2024 , publisher=

    2024

  39. [47]

    Bioinformatics , volume=

    AutoPCR: automated phenotype concept recognition by prompting , author=. Bioinformatics , volume=. 2026 , publisher=

    2026

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.