Pith. sign in

REVIEW 4 major objections 5 minor 51 references

Effective Multi-Task Learning for Biomedical Named Entity Recognition

T0 review · 4 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 SRU-NER, a transition-based named entity recognizer with a slot-based recurrent unit, matches near-state-of-the-art F1 on CoNLL-2003 and GENIA while improving cross-domain generalization through a multi-task loss that soft-targets missing…

desk verdict A credible nested-NER architecture with public code and strong single-task numbers, but the soft-target multi-task loss is under-specified and unablated, so the 'effective MTL' claim is not yet pinned down. read the letter →

arxiv 2507.18542 v1 pith:SLESBXBL submitted 2025-07-24 cs.CL

João Ruano , Gonçalo M. Correia , Leonor Barreiros , Afonso Mendes This is my paper · ORCID
classification cs.CL
keywords biomedicalnamedentityrecognitionnestedmulti-tasklearningtransition-basedparsingslot-basedrecurrentunitsofttargetsannotationgapscross-domaingeneralization
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 sets out to show that a single named entity recognition model can be trained on several biomedical datasets with different and partially overlapping annotation schemes without suffering from the false negatives those gaps usually cause. SRU-NER, a transition-based parser that emits open/close actions for entity types and uses a Slot-based Recurrent Unit to remember past actions, is the vehicle. The key training move is a soft-target loss: for entity types absent from a dataset, the model's own current prediction is used as the target, so it is never penalized for finding entities the dataset simply did not annotate. The authors report competitive single-task F1 of 94.48 on CoNLL-2003 and 80.10 on GENIA, and cross-corpus and human evaluations indicating better out-of-domain precision than comparable single-task or multi-task models.

What carries the argument

The load-bearing mechanism is a transition-based action encoding combined with a Slot-based Recurrent Unit (SRU) and a soft-target multi-task loss. The model emits a sequence of actions: TR(e_i) opens a mention of type e_i, RE(e_i) closes the most recently opened mention of that type, SH advances the token pointer, and EOA ends the cycle; nested mentions are encoded by the order of these actions. The SRU maintains a state matrix initialized with the sentence's contextual embeddings, adds a weighted action embedding to the row pointed to by the current token index at each step, and reads an output embedding through an attention mechanism over its slots. The multi-task loss then sets the target for actions of entity types absent from a given dataset to the model's own sigmoid output, while keeping gold one-hot targets for the annotated types, so the model is not penalized for predicting entities the dataset never labeled.

What would settle it

Train SRU-NER on the same six biomedical datasets with the soft-target mechanism replaced by simply masking (zeroing) the loss for actions of absent entity types, and compare disjoint-evaluation F1 on the same test splits; if masking matches or beats the soft-target model, the central multi-task claim fails.

Watch

Extended reading notes

Core claim

The central claim is that SRU-NER achieves competitive performance in both flat and nested NER and improves cross-domain generalization by handling annotation gaps dynamically. In the six-dataset multi-task experiment, it reaches the best or second-best F1 in the disjoint evaluation compared with earlier multi-task models, with an average of 84.48 in the merged and 86.63 in the disjoint scenario; the average of its six single-task models is 87.45. In the cross-corpus setup, its average mention-level F1 is 68.70 versus 66.67 for the replicated baseline. A human evaluation adds that the multi-task model is on average 25.4% more precise than single-task models at out-of-domain entity recognition, despite lower recall on some types.

Load-bearing premise

The training loss treats the model's own current predictions for entity types missing from a dataset as the correct soft targets, so the claimed multi-task gains depend on this self-generated signal being stable and not reinforcing errors; the paper does not ablate this mechanism.

Editorial extensions

If this is right

  • A single shared network can jointly decode entity types from disjoint annotation schemes, removing the need for task-specific heads or post-hoc conflict resolution.
  • Training on multiple partially annotated corpora raises out-of-domain precision, with the human evaluation finding the multi-task model 25.4% more precise on average than single-task models.
  • The action-based formulation supports nested mentions through per-type open/close stacks, so the same architecture covers flat and nested biomedical NER.
  • The soft-target loss makes the model usable on a union of datasets without adding false-negative penalties, which is the standard failure mode of naive dataset merging.
  • The cross-corpus F1 of 68.70 beats the replicated baseline of 66.67, suggesting the approach transfers to unseen biomedical corpora annotated for different entity types.

Reading between the lines

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

  • The soft-target mechanism is effectively a self-distillation toward the model's own beliefs; a natural next experiment is ablating it against simple masking to isolate whether the gain comes from the loss or the architecture.
  • Because the method treats same-named entity types across datasets as disjoint, type unification remains a post-processing step; learning shared type embeddings across datasets might further improve the merged evaluation.
  • The approach extends beyond NER: any sequence labeling task with partially annotated label sets, such as relation or event extraction, could use the same soft-target loss to train one model from multiple schemas.
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

4 major / 5 minor

Summary. The paper introduces SRU-NER, a transition-based NER model with a Slot-based Recurrent Unit, designed to handle flat and nested named entities while enabling multi-task learning over datasets with disjoint or partially overlapping entity type annotations. The proposed multi-task strategy modifies the loss so that entity types absent from a dataset are not penalized, and it inserts an extra shift step when the model begins an unannotated entity type. The model is evaluated on CoNLL-2003, GENIA, a six-dataset biomedical ensemble, a cross-corpus setting, and a human-annotated cross-domain experiment, reporting competitive F1 scores and improved out-of-domain precision over a single-task baseline.

Significance. If the reported results are reproducible, SRU-NER is a useful contribution to nested NER and to multi-task learning with partially annotated biomedical corpora. The public code release and the replication of existing cross-corpus evaluation settings are strengths, and the synthetic BC5CDR experiment in Table 5 provides some evidence that the overall training strategy helps global prediction. However, the central multi-task mechanism is under-specified: the self-referential soft targets for absent entity types are not described with respect to stop-gradient behavior and are not ablated, and the human evaluation's 'simulated recall' is not a true recall measure because it is computed over the union of the models' own predictions. These issues currently leave the paper's headline contribution underdetermined.

major comments (4)
  1. [Section 4] In the loss construction, G(t)_ai is set to sigma(u(t)_ai) for actions of entity types not annotated in the current dataset, but the paper does not state whether this target is detached from the computation graph. If it is detached, the binary cross-entropy gradient for those actions is identically zero, making the mechanism equivalent to hard masking absent types apart from the inserted SH row, which is itself unablated. If it is not detached, the loss term becomes a binary-entropy regularizer that drives the logits of absent types away from zero, a very different objective. The paper provides no implementation detail or ablation distinguishing these alternatives, so Tables 1, 4, and 6 cannot be used to attribute the gains to the proposed 'effective multi-task learning strategy.'
  2. [Appendix D / Table 6] The 'simulated recall' score is computed as the fraction of correct spans found by at least one of the three models, using the union of the models' predictions as the denominator, rather than gold annotations. This is not a recall measure: it conflates the candidate pool with the ground truth and makes recall values in Table 6 incomparable across models and entity types. The paper's claim that the multi-task model improves cross-domain generalization is therefore not supported by this experiment; the authors should either estimate recall on a manually annotated gold sample or restrict their claims to precision and coverage.
  3. [Section 5.1 / Tables 2 and 3] All F1 scores are reported as point estimates without variance, confidence intervals, or significance tests. For example, Table 2 shows SRU-NER at 94.48 versus 94.6 on CoNLL-2003 and 80.10 versus 81.53 on GENIA; these differences are well within typical run-to-run variability for neural NER systems. The competitive-performance claim requires at least multiple runs with standard deviations or bootstrap confidence intervals for the main comparisons, including Tables 1 and 4.
  4. [Table 1 and Section 5.2] Several comparison systems in Table 1 are marked with an asterisk indicating that they were trained on both training and development splits, whereas SRU-NER is trained on the training split only with early stopping on the development set. Section 5.1 acknowledges this difference for the single-task results, but Section 5.2's statement that the model achieves the 'best or second-best' scores in the disjoint evaluation does not qualify the comparison. To support the multi-task performance claim, the authors should either train SRU-NER under the same data conditions or explicitly state the data-availability disadvantage for each comparison.
minor comments (5)
  1. [Section 3.2] The word 'refered' should be 'referred'.
  2. [Section 3.1] In the action encoding description, 'RE(ej) appears after the SH representing the k-th word' appears to be a typo; it should likely read 'RE(ek) appears after the SH representing the j-th word,' since the mention starts at wi and ends at wj.
  3. [Section 4] The notation A_bE is used without an explicit definition; it should be introduced as the action set over the disjoint entity type union bE.
  4. [Appendix D] When 'simulated recall' is first mentioned in the main text, it should be clearly defined as a non-standard metric based on the union of model predictions, not gold annotations, to avoid misleading readers.
  5. [Table 12 / Appendix B] The sensitivity of the results to the latent embedding multiplier J and the token scaling parameter alpha is not reported; since these are architecture-level hyperparameters, at least a sentence or a small sensitivity check would be useful.

Circularity Check

1 steps flagged · score 4.0 of 10

Soft-target loss for absent entity types uses the model's own output as its training target; benchmark F1s remain externally grounded.

  1. self definitional [Section 4, 'Multi-task training strategy' (construction of G and loss L(t))]
    "More concretely, at time step t of the cycle, for ai∈A bE\A Ei, G(t) ai is set to be equal to σ(u(t) ai), where σ is the sigmoid function. ... This procedure ensures that G still reflects the original gold annotations in the columns corresponding to TR() and RE() actions of entity types in the source dataset, but incorporates the model’s probabilities for other actions."

    The target G for every absent-type action is defined as the model's own current sigmoid logit, so the BCE term over those coordinates contains no independent annotation signal. If the target is detached, the gradient is exactly σ(u)−G = 0, making the mechanism equivalent to hard-masking absent types; if it is not detached, the term minimizes the binary entropy of the model's own logits, driving confidence on types never annotated in that sentence. Either way, the 'effective multi-task learning' signal for annotation gaps is the model reinforcing its own output rather than a supervised cross-dataset signal.

full rationale

The single-task F1 scores (94.48 CoNLL-2003, 80.10 GENIA), multi-task results in Tables 1 and 4, and the synthetic BC5CDR experiment are all evaluated against external or gold/human-validated data, so the main performance claims are not circular. The one genuine self-referential step is the multi-task loss itself: for actions of types absent from a sentence's annotation, the paper sets the target G to the model's own σ(u), which removes annotation-derived supervision by construction (or, without stopping gradients, turns the term into an entropy minimizer). This step is load-bearing for contribution (ii), 'an effective multi-task training strategy', and it is unablated; the Limitations section acknowledges missing hyperparameter search but not this identification problem. Because the benchmark numbers have independent content, the score is moderate rather than high.

Assumptions & free parameters 4 free parameters · 4 assumptions · 1 invented entities

This is an empirical machine learning paper, so the ledger captures the hand-set thresholds and architectural assumptions that the NER performance depends on. The most load-bearing are the 0.5 decision thresholds, the SRU memory design, and the self-referential soft-target loss for absent entity types.

free parameters (4)
  • Action probability threshold = 0.5
    TR/RE actions are accepted only if sigmoid probability > 0.5 (Section 3.2, Decoding). This threshold is chosen by hand and directly determines decoded spans.
  • EOA stopping threshold = 0.5
    Action generation stops when sigmoid(u_EOA) > 0.5 (Equation 4). Hand-set.
  • SRU latent embedding multiplier J/|A_E| = 2 or 10
    Number of SRU latent embeddings is a multiple of |A_E|; set to 10 for GENIA single-task and 2 for all others (Table 12). Hand-chosen.
  • Token scaling parameter alpha = trained for GENIA, frozen to 1 otherwise
    Equations (5). The scaling of positional embeddings is trained only for the GENIA single-task model; in all other experiments it is frozen at 1 (Appendix B).
assumptions (4)
  • domain assumption Pretrained transformer encoders (LinkBERT-large, XLM-RoBERTa-large) provide contextual embeddings that transfer to biomedical NER.
    All models use LinkBERT-large or xlm-roberta-large as the encoder (Appendix B). The performance claims assume these pretrained representations are appropriate.
  • domain assumption The TR/RE action ordering with longest-first and shortest-first conventions is sufficient to encode all nested mention structures.
    Section 3.1 defines the action encoding and assumes any nested mentions can be represented by this deterministic ordering. No proof of completeness is given.
  • ad hoc to paper The Slot-based Recurrent Unit with Q=N+2 slots and attention over latent embeddings can accurately model the state of a parsed action history.
    Sections 3.2-3.3 introduce the SRU as the action-history memory. Its sufficiency is not proven; it is empirically motivated.
  • ad hoc to paper For entity types absent from a dataset, using the model's own sigmoid probability as the target in the loss is a valid training signal.
    Section 4 sets G(t)_ai = σ(u(t)_ai) for types not annotated in the source dataset. This is a design choice that is not ablated.
invented entities (1)
  • Slot-based Recurrent Unit (SRU) module independent evidence
    purpose: Stateful memory that stores weighted action embeddings in token-position-indexed slots and produces an attentional output embedding for action prediction.
    The SRU is a new architectural component. Its effectiveness is evidenced by the model's F1 scores on held-out benchmarks (Tables 1-3), providing a falsifiable handle outside the paper's own claims.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Effective Multi-Task Learning for Biomedical Named Entity Recognition." pith.science (2026). https://pith.science/paper/SLESBXBL

@misc{pith2026250718542,
  author       = {Pith},
  title        = {Pith review of: Effective Multi-Task Learning for Biomedical Named Entity Recognition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SLESBXBL}},
  note         = {Machine review of arXiv:2507.18542}
}
read the original abstract

Biomedical Named Entity Recognition presents significant challenges due to the complexity of biomedical terminology and inconsistencies in annotation across datasets. This paper introduces SRU-NER (Slot-based Recurrent Unit NER), a novel approach designed to handle nested named entities while integrating multiple datasets through an effective multi-task learning strategy. SRU-NER mitigates annotation gaps by dynamically adjusting loss computation to avoid penalizing predictions of entity types absent in a given dataset. Through extensive experiments, including a cross-corpus evaluation and human assessment of the model's predictions, SRU-NER achieves competitive performance in biomedical and general-domain NER tasks, while improving cross-domain generalization.

Figures

Figures reproduced from arXiv: 2507.18542 by the authors.

Figure 1
Figure 1. Action selection process for the sentence given in section 3.1, at time step t = 9. The gold nested mentions are "NF - chi B site", "chi B", of type DNA (D), and "NF - chi B" of type Protein (P). To compute the logits u (9), the model leverages the logits of the previous time steps, action embeddings and word embeddings. linguistic patterns. Recently, Transformer-based architectures have set new benchmarks, driving … view at source ↗
Figure 2
Figure 2. SRU unit at time step t. Its internal state is updated depending on its current state C(t) and the weighted action embeddings Ω (t) . This stateful function also leverages a set of latent representations. It produces an output embedding h (t+1) by applying an attention mechanism to the updated state. p (t) ∈ {0, 1, . . . , Q − 1} is an input integer. It also produces an output embedding h (t+1) ∈ R d via h (t+1) = g… view at source ↗
Figure 3
Figure 3. Example of a sentence from the test split of the BC5CDR corpus (Li et al., 2016), together with gold spans and predicted spans as annotated by the MTL model described in section 5.2. The model is trained on six datasets, covering eight entity types ∪i Ei = {Chemical, Disease, . . .}. Notice that some of these types are common to multiple datasets (namely, ’Chemical’, annotated on both the BC4CHEMD and BC5CDR dataset… view at source ↗

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 24 canonical work pages

  1. [1]

    Cecilia Arighi, Lynette Hirschman, Thomas Lemberger, Samuel Bayer, Robin Liechti, Donald Comeau, and Cathy Wu. 2017. Bio-id track overview. In BioCreative VI Challenge Evaluation Workshop, volume 482, page 376

    work page 2017

  2. [2]

    Pratyay Banerjee, Kuntal Kumar Pal, Murthy Devarakonda, and Chitta Baral. 2021. https://doi.org/10.1145/3465221 Biomedical Named Entity Recognition via Knowledge Guidance and Question Answering . ACM Trans. Comput. Healthcare, 2(4):33:1--33:24

    work page doi:10.1145/3465221 2021

  3. [3]

    Nigel Collier, Tomoko Ohta, Yoshimasa Tsuruoka, Yuka Tateisi, and Jin-Dong Kim. 2004. https://aclanthology.org/W04-1213/ Introduction to the bio-entity recognition task at JNLPBA . In Proceedings of the International Joint Workshop on Natural Language Processing in Biomedicine and its Applications ( NLPBA / B io NLP ) , pages 73--78, Geneva, Switzerland. COLING

    work page 2004

  4. [4]

    Alexis Conneau, Kartikay Khandelwal, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzm \'a n, Edouard Grave, Myle Ott, Luke Zettlemoyer, and Veselin Stoyanov. 2020. https://doi.org/10.18653/v1/2020.acl-main.747 Unsupervised cross-lingual representation learning at scale . In Proceedings of the 58th Annual Meeting of the Association for Comp...

    doi:10.18653/v1/2020.acl-main.747 2020

  5. [5]

    Gamal Crichton, Sampo Pyysalo, Billy Chiu, and Anna Korhonen. 2017. https://doi.org/10.1186/s12859-017-1776-8 A neural network multi-task learning approach to biomedical named entity recognition . BMC Bioinformatics, 18(1):368

    doi:10.1186/s12859-017-1776-8 2017

  6. [6]

    Rezarta Islamaj Do g an, Robert Leaman, and Zhiyong Lu. 2014. NCBI disease corpus: a resource for disease name recognition and concept normalization. J Biomed Inform, 47:1--10

    work page 2014

  7. [7]

    Chris Dyer, Miguel Ballesteros, Wang Ling, Austin Matthews, and Noah A. Smith. 2015. https://doi.org/10.3115/v1/P15-1033 Transition-based dependency parsing with stack long short-term memory . In Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing...

    doi:10.3115/v1/p15-1033 2015

  8. [8]

    Octavian-Eugen Ganea and Thomas Hofmann. 2017. https://doi.org/10.18653/v1/D17-1277 Deep joint entity disambiguation with local neural attention . In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 2619--2629, Copenhagen, Denmark. Association for Computational Linguistics

    doi:10.18653/v1/d17-1277 2017

Show all 51 references
  1. [9]

    Martin Gerner, Goran Nenadic, and Casey M. Bergman. 2010. https://doi.org/10.1186/1471-2105-11-85 Linnaeus: A species name identification system for biomedical literature . BMC Bioinformatics, 11(1):85

    2010 doi

  2. [10]

    Nathan Greenberg, Trapit Bansal, Patrick Verga, and Andrew McCallum. 2018. https://doi.org/10.18653/v1/D18-1306 Marginal likelihood training of B i LSTM - CRF for biomedical named entity recognition from disjoint label sets . In Proceedings of the 2018 Conference on Empirical ...

    2018 doi

  3. [11]

    Harsha Gurulingappa, Roman Klinger, Martin Hofmann-Apitius, and Juliane Fluck. 2010. http://www.nactem.ac.uk/biotxtm/papers/Gurulingappa.pdf An empirical evaluation of resources for the identification of diseases and adverse effects in biomedical literature . In 2nd Workshop o...

    2010

  4. [12]

    Maryam Habibi, Leon Weber, Mariana Neves, David Luis Wiegandt, and Ulf Leser. 2017. https://doi.org/10.1093/bioinformatics/btx228 Deep learning with word embeddings improves biomedical named entity recognition . Bioinformatics, 33(14):i37--i48

    2017 doi

  5. [13]

    Xiao Huang, Li Dong, Elizabeth Boschee, and Nanyun Peng. 2019. https://doi.org/10.18653/v1/K19-1048 Learning a unified named entity tagger from multiple partially annotated corpora for efficient adaptation . In Proceedings of the 23rd Conference on Computational Natural Langua...

    2019 doi

  6. [14]

    Rezarta Islamaj, Robert Leaman, Sun Kim, Dongseop Kwon, Chih-Hsuan Wei, Donald C Comeau, Yifan Peng, David Cissel, Cathleen Coss, Carol Fisher, Rob Guzman, Preeti Gokal Kochar, Stella Koppel, Dorothy Trinh, Keiko Sekiya, Janice Ward, Deborah Whitman, Susan Schmidt, and Zhiyong...

    2021

  7. [15]

    Rezarta Islamaj, Chih-Hsuan Wei, David Cissel, Nicholas Miliaras, Olga Printseva, Oleg Rodionov, Keiko Sekiya, Janice Ward, and Zhiyong Lu. 2021 b . NLM-Gene , a richly annotated gold standard dataset for gene entities that addresses ambiguity and multi-species gene recognitio...

    2021

  8. [16]

    Imed Keraghel, Stanislas Morbieu, and Mohamed Nadif. 2024. https://arxiv.org/abs/2401.10825 Recent advances in named entity recognition: A comprehensive survey and comparative study . Preprint, arXiv:2401.10825

    2024 arXiv

  9. [17]

    Muhammad Raza Khan, Morteza Ziyadi, and Mohamed AbdelHady. 2020. https://arxiv.org/abs/2001.08904 Mt-bioner: Multi-task learning for biomedical named entity recognition using deep bidirectional transformers . Preprint, arXiv:2001.08904

    2020 arXiv

  10. [18]

    J.-D. Kim, T. Ohta, Y. Tateisi, and J. Tsujii. 2003. https://doi.org/10.1093/bioinformatics/btg1023 Genia corpus—a semantically annotated corpus for bio-textmining . Bioinformatics, 19:i180--i182

    2003 doi

  11. [19]

    Friedrich, Martin Hofmann-Apitius, and Juliane Fluck

    Corinna Kolarik, Roman Klinger, Christoph M. Friedrich, Martin Hofmann-Apitius, and Juliane Fluck. 2008. http://www.romanklinger.de/publications/kolarik2008.pdf Chemical Names: Terminological Resources and Corpora Annotation . In Workshop on Building and evaluating resources f...

    2008

  12. [20]

    Lowe, Roger A

    Martin Krallinger, Obdulia Rabal, Florian Leitner, Miguel Vazquez, David Salgado, Zhiyong Lu, Robert Leaman, Yanan Lu, Donghong Ji, Daniel M. Lowe, Roger A. Sayle, Riza Theresa Batista-Navarro, Rafal Rak, Torsten Huber, Tim Rockt \"a schel, S \'e rgio Matos, David Campos, Buzh...

    2015

  13. [21]

    Jiao Li, Yueping Sun, Robin J Johnson, Daniela Sciaky, Chih-Hsuan Wei, Robert Leaman, Allan Peter Davis, Carolyn J Mattingly, Thomas C Wiegers, and Zhiyong Lu. 2016. BioCreative V CDR task corpus: a resource for chemical disease relation extraction. Database (Oxford), 2016

    2016

  14. [22]

    Jing Li, Aixin Sun, Jianglei Han, and Chenliang Li. 2022. https://doi.org/10.1109/TKDE.2020.2981314 A survey on deep learning for named entity recognition . IEEE Transactions on Knowledge and Data Engineering, 34(1):50--70

    2022

  15. [23]

    Ling Luo, Po-Ting Lai, Chih-Hsuan Wei, Cecilia N Arighi, and Zhiyong Lu. 2022. https://doi.org/10.1093/bib/bbac282 Biored: a rich biomedical relation extraction dataset . Briefings in Bioinformatics, 23(5):bbac282

    2022 doi

  16. [24]

    Ling Luo, Chih-Hsuan Wei, Po-Ting Lai, Robert Leaman, Qingyu Chen, and Zhiyong Lu. 2023. https://doi.org/10.1093/bioinformatics/btad310 AIONER: all-in-one scheme-based biomedical named entity recognition using deep learning . Bioinformatics, 39(5):btad310

    2023 doi

  17. [25]

    Zita Marinho, Afonso Mendes, Sebasti \ a o Miranda, and David Nogueira. 2019. https://doi.org/10.18653/v1/W19-1904 Hierarchical nested named entity recognition . In Proceedings of the 2nd Clinical Natural Language Processing Workshop, pages 28--34, Minneapolis, Minnesota, USA....

    2019 doi

  18. [26]

    Tahir Mehmood, Alfonso Gerevini, Alberto Lavelli, and Ivan Serina. 2019. Leveraging multi-task learning for biomedical named entity recognition. In AI*IA 2019 -- Advances in Artificial Intelligence, pages 431--444, Cham. Springer International Publishing

    2019

  19. [27]

    Sunil Mohan and Donghui Li. 2019. https://arxiv.org/abs/1902.09476 Medmentions: A large biomedical corpus annotated with umls concepts . Preprint, arXiv:1902.09476

    2019 arXiv

  20. [28]

    Vincenzo Moscato, Marco Postiglione, Carlo Sansone, and Giancarlo Sperlí. 2023. https://doi.org/10.1109/JBHI.2023.3244044 Taughtnet: Learning multi-task biomedical named entity recognition from single-task teachers . IEEE Journal of Biomedical and Health Informatics, 27(5):2512--2523

    2023

  21. [29]

    Evangelos Pafilis, Sune P Frankild, Lucia Fanini, Sarah Faulwetter, Christina Pavloudi, Aikaterini Vasileiadou, Christos Arvanitidis, and Lars Juhl Jensen. 2013. The SPECIES and ORGANISMS resources for fast and accurate identification of taxonomic names in text. PLoS One, 8(6):e65390

    2013

  22. [30]

    Yesol Park, Gyujin Son, and Mina Rho. 2024. https://doi.org/10.3390/app14209302 Biomedical flat and nested named entity recognition: Methods, challenges, and advances . Applied Sciences, 14(20)

    2024 doi

  23. [31]

    Abhishek Sharma, Amrita, Sudeshna Chakraborty, and Shivam Kumar. 2022. Named entity recognition in natural language processing: A systematic review. In Proceedings of Second Doctoral Symposium on Computational Intelligence, pages 817--828, Singapore. Springer Singapore

    2022

  24. [32]

    Yongliang Shen, Xinyin Ma, Zeqi Tan, Shuai Zhang, Wen Wang, and Weiming Lu. 2021. https://doi.org/10.18653/v1/2021.acl-long.216 Locate and label: A two-stage identifier for nested named entity recognition . In Proceedings of the 59th Annual Meeting of the Association for Compu...

    2021 doi

  25. [33]

    Yongliang Shen, Kaitao Song, Xu Tan, Dongsheng Li, Weiming Lu, and Yueting Zhuang. 2023. https://doi.org/10.18653/v1/2023.acl-long.215 D iffusion NER : Boundary diffusion for named entity recognition . In Proceedings of the 61st Annual Meeting of the Association for Computatio...

    2023 doi

  26. [34]

    Yongliang Shen, Xiaobin Wang, Zeqi Tan, Guangwei Xu, Pengjun Xie, Fei Huang, Weiming Lu, and Yueting Zhuang. 2022. https://doi.org/10.18653/v1/2022.acl-long.67 Parallel instance query network for named entity recognition . In Proceedings of the 60th Annual Meeting of the Assoc...

    2022 doi

  27. [35]

    Tanabe, Rie Johnson nee Ando, Cheng-Ju Kuo, I-Fang Chung, Chun-Nan Hsu, Yu-Shi Lin, Roman Klinger, Christoph M

    Larry Smith, Lorraine K. Tanabe, Rie Johnson nee Ando, Cheng-Ju Kuo, I-Fang Chung, Chun-Nan Hsu, Yu-Shi Lin, Roman Klinger, Christoph M. Friedrich, Kuzman Ganchev, Manabu Torii, Hongfang Liu, Barry Haddow, Craig A. Struble, Richard J. Povinelli, Andreas Vlachos, William A. Bau...

    2008

  28. [36]

    Cong Sun, Zhihao Yang, Lei Wang, Yin Zhang, Hongfei Lin, and Jian Wang. 2021. https://www.sciencedirect.com/science/article/pii/S1532046421001283 Biomedical named entity recognition using bert in the machine reading comprehension framework . Journal of Biomedical Informatics, ...

    2021

  29. [37]

    Mario Sänger, Samuele Garda, Xing David Wang, Leon Weber-Genzel, Pia Droop, Benedikt Fuchs, Alan Akbik, and Ulf Leser. 2024. https://doi.org/10.1093/bioinformatics/btae564 Hunflair2 in a cross-corpus evaluation of biomedical named entity recognition and normalization tools . B...

    2024 doi

  30. [38]

    Zeqi Tan, Yongliang Shen, Shuai Zhang, Weiming Lu, and Yueting Zhuang. 2021. https://doi.org/10.24963/ijcai.2021/542 A sequence-to-set network for nested named entity recognition . In Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence, IJCAI...

    2021 doi

  31. [39]

    Tjong Kim Sang and Fien De Meulder

    Erik F. Tjong Kim Sang and Fien De Meulder. 2003. https://aclanthology.org/W03-0419/ Introduction to the C o NLL -2003 shared task: Language-independent named entity recognition . In Proceedings of the Seventh Conference on Natural Language Learning at HLT - NAACL 2003 , pages...

    2003

  32. [40]

    Xinyu Wang, Yong Jiang, Nguyen Bach, Tao Wang, Zhongqiang Huang, Fei Huang, and Kewei Tu. 2021. https://doi.org/10.18653/v1/2021.acl-long.206 Automated concatenation of embeddings for structured prediction . In Proceedings of the 59th Annual Meeting of the Association for Comp...

    2021 doi

  33. [41]

    Xuan Wang, Yu Zhang, Xiang Ren, Yuhao Zhang, Marinka Zitnik, Jingbo Shang, Curtis Langlotz, and Jiawei Han. 2018. https://doi.org/10.1093/bioinformatics/bty869 Cross-type biomedical named entity recognition with deep multi-task learning . Bioinformatics, 35(10):1745--1752

    2018 doi

  34. [42]

    Chih-Hsuan Wei, Alexis Allot, Kevin Riehle, Aleksandar Milosavljevic, and Zhiyong Lu. 2022. https://doi.org/10.1093/bioinformatics/btac537 tmvar 3.0: an improved variant concept recognition and normalization tool . Bioinformatics, 38(18):4449--4451

    2022 doi

  35. [43]

    Chih-Hsuan Wei, Hung-Yu Kao, and Zhiyong Lu. 2015. GNormPlus : An integrative approach for tagging genes, gene families, and protein domains. Biomed Res Int, 2015:918710

    2015

  36. [44]

    Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pierric Cistac, Tim Rault, Remi Louf, Morgan Funtowicz, Joe Davison, Sam Shleifer, Patrick von Platen, Clara Ma, Yacine Jernite, Julien Plu, Canwen Xu, Teven Le Scao, Sylvain Gugger, Mari...

    2020 doi

  37. [45]

    Hang Yan, Yu Sun, Xiaonan Li, and Xipeng Qiu. 2023. https://doi.org/10.18653/v1/2023.acl-short.123 An embarrassingly easy but strong baseline for nested named entity recognition . In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volum...

    2023 doi

  38. [46]

    Michihiro Yasunaga, Jure Leskovec, and Percy Liang. 2022. https://doi.org/10.18653/v1/2022.acl-long.551 L ink BERT : Pretraining language models with document links . In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Pap...

    2022 doi

  39. [47]

    Yu Yin, Hyunjae Kim, Xiao Xiao, Chih Hsuan Wei, Jaewoo Kang, Zhiyong Lu, Hua Xu, Meng Fang, and Qingyu Chen. 2024. https://doi.org/10.1016/j.jbi.2024.104731 Augmenting biomedical named entity recognition with general-domain resources . Journal of Biomedical Informatics, 159:104731

    2024

  40. [48]

    Juntao Yu, Bernd Bohnet, and Massimo Poesio. 2020. https://doi.org/10.18653/v1/2020.acl-main.577 Named entity recognition as dependency parsing . In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 6470--6476, Online. Association f...

    2020 doi

  41. [49]

    Sheng Zhang, Hao Cheng, Jianfeng Gao, and Hoifung Poon. 2023. https://openreview.net/forum?id=9EAQVEINuum Optimizing bi-encoder for named entity recognition via contrastive learning . In The Eleventh International Conference on Learning Representations

    2023

  42. [50]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list...

  43. [51]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

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