Pith. sign in

REVIEW 5 major objections 4 minor 42 references

Multi-Relation Extraction in Entity Pairs using Global Context

T0 review · 5 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Appending entity text to the document yields state-of-the-art relation extraction on DocRED.

desk verdict The input encoding is a plausible minor variant, but the reported SOTA results are invalid because the paper compares dev-set scores and a different training split against published test-set numbers, so the central claim collapses until the evaluation is redone on official splits. read the letter →

arxiv 2507.22926 v1 pith:G6YZCKMW submitted 2025-07-23 cs.CL cs.IR

Nilesh , Atul Gupta , Avinash C Panday This is my paper
classification cs.CLcs.IR
keywords document-levelrelationextractionglobalcontextinputencodingBERTmulti-sentencereasoningDocREDRe-DocREBEL
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 proposes a document-level relation extraction method in which the input to BERT is the whole document followed by the head and tail entity texts, each placed in a separate segment. The [CLS] token representation is then classified into relation types, so the model must aggregate evidence from the entire document rather than only co-occurrence sentences. The authors report that this input encoding outperforms graph-based and transformer-based systems on DocRED and Re-DocRED, and also does well on the REBEL dataset, using a stock BERT encoder and a single classification layer. The significance would be that a simple representational change can capture multi-sentence reasoning without task-specific architecture.

What carries the argument

The input encoding strategy: the document tokens are placed after [CLS] with segment A, and after a [SEP] marker the head entity tokens and tail entity tokens are appended in segment B with their own [SEP] markers. Positional embeddings run across the whole sequence, so the transformer can relate any document token to the entity pair. The classification layer takes the pooled [CLS] output through a tanh, dropout, and a linear-softmax head, and the model is trained with a cross-entropy loss over relation types. This mechanism works by forcing the representation to be conditioned on the entity pair while still attending to the full document, so evidence from distant sentences can contribute to the prediction.

What would settle it

Run the proposed method on the official DocRED test split (the 1,000-document held-out set) and on Re-DocRED using its own training split; if the reported 88.91% and 67.19% F1 numbers do not reproduce, the central claim of state-of-the-art performance collapses.

Watch

Extended reading notes

Core claim

The central claim is that representing an entity pair as appended text segments to the document sequence gives a pretrained transformer all the global context it needs for document-level relation extraction. The paper shows that the [CLS] token, after encoding the concatenated sequence, can be mapped through a softmax to predict the relation between the head and tail entities, and that this works across three datasets. The authors state that their method achieved higher precision, recall, and F1 than existing methods on the validation and test sets of all considered datasets, with the largest reported gap on DocRED (88.91% test F1 versus 66.31% for the previous best).

Load-bearing premise

The reported advantage over existing systems assumes that the F1 numbers in the comparison tables are measured under the same protocol as prior work; the paper labels the DocRED dev set as 'test' and trains on DocRED's distantly supervised split for the Re-DocRED experiment, so the comparison may not be apples-to-apples.

Editorial extensions

If this is right

  • On DocRED, the reported test F1 of 88.91% would place the method above all systems listed in the comparison table, including DocRE-CLiP's 66.31%.
  • The approach classifies each entity pair individually, so it can handle any number of relation types without changing the architecture, as the classification layer is adapted to the dataset's relation set.
  • The same input encoding transfers to a sentence-level relation extraction dataset (REBEL) without modification, suggesting it is not tied to document-level specifics.
  • If the reported numbers are taken at face value, the method offers a simpler alternative to graph-based and contrastive-learning DocRE systems while performing at least as well.

Reading between the lines

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

  • Correcting the evaluation protocol—using the official DocRED test split and Re-DocRED's own training split—could substantially reduce the reported margins; a fair comparison would likely show the method competitive rather than state-of-the-art.
  • The encoding is complementary to other DocRE components such as graph networks or adaptive thresholding, so combining them might yield further gains.
  • Because the model treats each entity pair independently, it ignores inter-pair dependencies that some recent systems exploit; a shared representation across pairs could improve consistency.
  • The unusually high REBEL score suggests the benchmark or the split may not be directly comparable to prior REBEL evaluations; a sanity check on the dataset statistics would clarify.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 4 minor

Summary. The paper proposes a document-level relation extraction method built on BERT. The input sequence concatenates the full document text with the head entity and tail entity texts as separate segments, and the [CLS] representation is passed through a fully connected layer followed by softmax to predict the relation for a given entity pair. The authors claim state-of-the-art or better F1 results on DocRED, Re-DocRED, and REBEL, reporting large margins over existing systems in Tables V and VI. The contribution is primarily empirical: a new input encoding for global context, with experiments on three datasets and comparisons to published baselines.

Significance. If the reported results were valid, the proposed encoding would be a surprisingly simple and effective way to inject global context into BERT for document-level relation extraction, and the cross-dataset results on DocRED, Re-DocRED, and REBEL would be noteworthy. The paper does not release code, and the empirical evaluation is the entire basis of the contribution, so the correctness of the evaluation protocol is load-bearing. Several visible protocol problems, however, invalidate the central comparison, and the single-label classification head is mismatched with the multi-label nature of the benchmarks. The paper's own tables and text contain internal contradictions about whether the method surpasses or merely approaches baselines on Re-DocRED. I therefore do not think the central claim is currently supported.

major comments (5)
  1. [IV-C, Tables II and V] The reported DocRED test F1 of 88.91% is not computed on the official DocRED test set. Table II is headed "Train & Val: DocRED train distant Test: DocRED (Dev set)", and Table V labels the 88.91 value as "Test F1". DocRED's official test set is held out and evaluated through the CodaLab server; evaluating on the dev set and calling it test F1 is a different protocol. The comparison against published test-set numbers in Table V is therefore invalid, and the claimed 22.60-point improvement over DocRE-CLiP is not supported.
  2. [IV-C, Tables III and VI] The Re-DocRED result is also obtained under a non-comparable protocol. Table III is headed "Train & Val: DocRED train distant Test: Re-DocRED (Test set)", meaning the model was trained on DocRED distant supervision rather than on Re-DocRED's own training split. The baselines in Table VI train on Re-DocRED's training data, so the reported 67.19 test F1 cannot be compared with them. Additionally, even under this mismatched protocol, 67.19 is not higher than the DREEAM test F1 of 67.53 reported in the same table, which contradicts the text claiming that the proposed method surpasses other approaches by a significant margin.
  3. [III, Algorithm 1, lines 11-12] The classification layer applies softmax followed by argmax over relations, which selects exactly one relation per entity pair. DocRED and Re-DocRED are multi-label datasets: an entity pair can hold several valid relations simultaneously. A single-label output head cannot reproduce the multi-label evaluation setting of the baselines without a thresholding or multi-label classification head. This mismatch means the reported precision, recall, and F1 values are not measuring the same task as the cited systems, independent of the data-split issues.
  4. [IV-A, REBEL paragraph and Table IV] The REBEL dataset is not adequately identified and is not evaluated against any baseline. The text describes REBEL as a BART-based seq2seq relation-extraction dataset, citing reference [28], but the cited NeurIPS 2024 paper is about reinforcement learning via regressing relative rewards, not the REBEL relation-extraction model/dataset. Table IV reports roughly 93% F1 on REBEL with no comparison to prior work, so it cannot support the conclusion that the proposed method outperforms state-of-the-art methods on this dataset.
  5. [V, Conclusions] The conclusion contains an internal contradiction. The introduction and abstract state that the method achieves higher precision, recall, and F1 on all considered datasets, but the concluding paragraph says "its performance on the Re-DocRED dataset still requires enhancement" and lists false negatives on Re-DocRED as future work. Since Table VI shows a Re-DocRED test F1 below the best baseline, the blanket claim of superiority on all datasets is not consistent with the paper's own reported numbers.
minor comments (4)
  1. [IV-B] The experimental setup states that 80% of the data was used for training and the remainder for testing and validation, but DocRED and Re-DocRED have fixed official train/dev/test splits. The relationship between this custom split and the official splits used in Tables II and III is unclear and should be stated precisely.
  2. [IV-C, Figure 4] Figure 4 is captioned "Training and Validation Loss-10 Epochs", but the text says the optimal number of epochs is three and all reported tables show only three epochs. Please clarify whether training was run for 10 epochs and only three are shown, or whether the caption is incorrect.
  3. [IV-C, Re-DocRED comparison paragraph] The sentence "indicating a 24.61% increase on the validation set and comparable performance on the test set" is confusing: 84.00 vs. 67.41 is a 24.6% relative increase, but this is not an absolute percentage-point gap, and the test values are not comparable because of the protocol mismatch.
  4. [Throughout] There are numerous typographical and formatting errors, including "avish.p@iiitdmj.ac.in" with a period before @, "DRN [31]„" with a stray comma, "BER T" in Tables V and VI, and an incomplete parenthesis in Section II in the sentence about multi-label classification. These should be corrected.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the proposed method is an empirical evaluation against external benchmarks, and the reported protocol issues concern benchmarking validity, not derivation circularity.

full rationale

The paper contains no derivation chain that folds its target into its own assumptions. The method is a direct empirical setup: it defines an input encoding (concatenating document tokens, entity segments, [CLS], and [SEP]), feeds the [CLS] representation through a softmax classifier, and trains on the labels of DocRED, Re-DocRED, and REBEL. No equation in the paper defines the predicted relation in terms of the reported F1 scores, and no fitted parameter is renamed as a prediction. There are no load-bearing self-citations, no imported uniqueness theorems, and no ansatz smuggled in through prior work by the authors; the cited baselines are external systems. The identified benchmarking concerns, such as Table II labeling the DocRED dev set as 'test' and Table III reporting a Re-DocRED test result from a model trained on DocRED distant data, are threats to the validity of the comparison, not circularity under the definitions used here. Similarly, the single-label softmax and argmax in Algorithm 1 lines 11-12 may be incompatible with multi-label evaluation, but that is a correctness or protocol issue rather than a case where the claim reduces to its inputs by construction. The central empirical claim is therefore self-contained in the sense required by this pass, and the appropriate finding is no significant circularity.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The method rests on standard BERT machinery plus a set of unstated domain assumptions about dataset protocols and task formulation. The key assumptions are that DocRED is a single-label task, that documents fit in the BERT window, and that cross-dataset evaluation is valid. None of these are defended.

free parameters (3)
  • Number of training epochs = 3
    Selected by inspecting validation loss and accuracy in Figure 4; the test numbers in Tables II-IV are reported for epoch 2, not epoch 3, which complicates the choice.
  • Dropout probability = 0.3
    Chosen by hand for the pooled [CLS] representation, stated in Section III-B.
  • Batch size = 16, 32, or 64 (unspecified)
    The paper says batch sizes 16, 32, and 64 were explored, but the final batch size used for the reported results is not stated.
assumptions (5)
  • standard math Transformer architecture and pretrained BERT weights from bert-base-uncased are valid and applicable without modification to document-level relation extraction.
    The method relies on BERT's self-attention and pooled [CLS] representation (Section III-B); no formal justification is given for applying it to long documents.
  • domain assumption Documents fit within BERT's 512-token input window, or truncation has no material effect.
    The input sequence concatenates the full document with two entity segments, but the paper never specifies a maximum length or truncation strategy.
  • domain assumption Relation extraction in DocRED and Re-DocRED can be treated as single-label classification.
    Algorithm 1 uses softmax and arg max over 96 classes, while DocRED allows multiple relations per entity pair; the paper does not address this mismatch.
  • ad hoc to paper Training on DocRED and evaluating on Re-DocRED is a valid measure of Re-DocRED performance.
    Table III trains on DocRED train distant and reports test F1 on Re-DocRED, while SOTA systems are trained on Re-DocRED's own split.
  • domain assumption The cited REBEL paper [28] describes the REBEL dataset used in the experiments.
    Reference [28] is actually a reinforcement learning paper, not the REBEL relation extraction dataset, so the dataset provenance is unverified.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multi-Relation Extraction in Entity Pairs using Global Context." pith.science (2026). https://pith.science/paper/G6YZCKMW

@misc{pith2026250722926,
  author       = {Pith},
  title        = {Pith review of: Multi-Relation Extraction in Entity Pairs using Global Context},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/G6YZCKMW}},
  note         = {Machine review of arXiv:2507.22926}
}
read the original abstract

In document-level relation extraction, entities may appear multiple times in a document, and their relationships can shift from one context to another. Accurate prediction of the relationship between two entities across an entire document requires building a global context spanning all relevant sentences. Previous approaches have focused only on the sentences where entities are mentioned, which fails to capture the complete document context necessary for accurate relation extraction. Therefore, this paper introduces a novel input embedding approach to capture the positions of mentioned entities throughout the document rather than focusing solely on the span where they appear. The proposed input encoding approach leverages global relationships and multi-sentence reasoning by representing entities as standalone segments, independent of their positions within the document. The performance of the proposed method has been tested on three benchmark relation extraction datasets, namely DocRED, Re-DocRED, and REBEL. The experimental results demonstrated that the proposed method accurately predicts relationships between entities in a document-level setting. The proposed research also has theoretical and practical implications. Theoretically, it advances global context modeling and multi-sentence reasoning in document-level relation extraction. Practically, it enhances relationship detection, enabling improved performance in real-world NLP applications requiring comprehensive entity-level insights and interpretability.

Figures

Figures reproduced from arXiv: 2507.22926 by the authors.

Figure 1
Figure 1. Example of DRE Document-level relation extraction (DRE) presents unique challenges, such as cross-sentence dependencies and the necessity for long-distance reasoning. In this context, im￾plicit connections require analyzing the entire document rather than isolated sentences. Despite recent advance￾ments, DRE remains complex due to referential entities and indirect connections. Therefore, this paper introduces a nove… view at source ↗
Figure 2
Figure 2. It combines embeddings for document tokens, [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 2
Figure 2. Input Encoding B. BERT Encoder layers The proposed method employs a pre-trained BERT (bert-base-uncased) without any additional refinement of its encoder, thereby preserving the original BERT weights and its inherent generalization capability. Rather than extracting the complete set of hidden states, the proposed method relies solely on the pooled output, specifically, the [CLS] token following a tanh activation on … view at source ↗
Figures from the paper (8 more)
Figure 3
Figure 3. Figure 3: Relation Classifier TABLE I Dataset Statistics Dataset #Doc. #Word #Sent. #Ent. #Rel. #Inst. #Fact DocRED (Human-annotated) 5053 1,002k 40276 132375 96 63427 56354 DocRED (Distantly Supervised) 101873 21,368k 828115 2558350 96 1508320 881298 Re-DocRED 5053 1.5M 92000 1…
Figure 4
Figure 4. Figure 4: Training and Validation Loss-10 Epochs the optimal number of epochs that avoids both underfit￾ting and overfitting. The figure shows that the loss and accuracy degrade after the third epoch, indicating that three epochs offer the best balance. The performance of the pr…
Figure 5
Figure 5. Figure 5: Precision, Recall, and F1 Score Curve Moreover, a precision-recall curve is also drawn in Fig￾ure 6 for the DocRed dataset to highlight the proposed method’s overall effectiveness. It can be observed from the [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 8
Figure 8. Figure 8: The proposed method predicts the appropriate [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 6
Figure 6. Figure 6: Precision and Recall Curve figure that the proposed method demonstrates strong per￾formance in identifying true positives while maintaining a high rate of accurate predictions. An increase in precision and recall indicates that the proposed method is becoming more accu…
Figure 7
Figure 7. Figure 7: Loss Comparision of evaluation matrics on three Dataset [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 8
Figure 8. Figure 8: Example of DRE. Note that mentions of the same entity are marked with identical colors. The output of the proposed method on [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]
Figure 9
Figure 9. Figure 9: Output Relations of the proposed method on same entity pair on different Local context [PITH_FULL_IMAGE:figures/full_fig_p010_9.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

42 extracted references · 36 canonical work pages

  1. [28]

    Rebel: Re- inforcement learning via regressing relative rewards,

    Z. Gao, J. Chang, W. Zhan, O. Oertell, G. Swamy, K. Brantley, T. Joachims, D. Bagnell, J. D. Lee, and W. Sun, “Rebel: Re- inforcement learning via regressing relative rewards,” Advances in Neural Information Processing Systems, vol. 37, pp. 52 354– 52 400, 2025

    work page 2025

  2. [1]

    Relation extraction: A survey,

    S. Pawar, G. K. Palshikar, and P. Bhattacharyya, “Relation extraction: A survey,” arXiv preprint arXiv:1712.05191, 2017

    arXiv 2017

  3. [2]

    A comprehensive survey of document-level relation extraction (2016-2023),

    J. Delaunay, H. T. H. Tran, C.-E. Gonz´ alez-Gallardo, G. Bor- dea, N. Sidere, and A. Doucet, “A comprehensive survey of document-level relation extraction (2016-2023),” arXiv preprint arXiv:2309.16396, 2023

    arXiv 2016

  4. [3]

    A comprehensive survey on relation extraction: Recent advances and new frontiers,

    X. Zhao, Y. Deng, M. Yang, L. Wang, R. Zhang, H. Cheng, W. Lam, Y. Shen, and R. Xu, “A comprehensive survey on relation extraction: Recent advances and new frontiers,” ACM Computing Surveys, vol. 56, no. 11, pp. 1–39, 2024. Fig. 7. Loss Comparision of evaluation matrics on three Dataset Fig. 8. Example of DRE. Note that mentions of the same entity are mar...

    work page 2024

  5. [4]

    Dual-channel and hierarchical graph convolu- tional networks for document-level relation extraction,

    Q. Sun, T. Xu, K. Zhang, K. Huang, L. Lv, X. Li, T. Zhang, and D. Dore-Natteh, “Dual-channel and hierarchical graph convolu- tional networks for document-level relation extraction,” Expert Systems with Applications, vol. 205, p. 117678, 2022

    work page 2022

  6. [5]

    Document- level relation extraction with global and path dependencies,

    W. Jia, R. Ma, L. Yan, W. Niu, and Z. Ma, “Document- level relation extraction with global and path dependencies,” Knowledge-Based Systems, vol. 289, p. 111545, 2024

    work page 2024

  7. [6]

    Corex: Document-level relation extraction framework with consistent two-hop reasoning and evidence sen- tence prediction,

    S. Zhao and C. Li, “Corex: Document-level relation extraction framework with consistent two-hop reasoning and evidence sen- tence prediction,” in 2024 International Joint Conference on Neural Networks (IJCNN). IEEE, 2024, pp. 1–8

    work page 2024

  8. [7]

    Inter-sentence Relation Extraction with Document-level Graph Convolutional Neural Network

    S. K. Sahu, F. Christopoulou, M. Miwa, and S. Ananiadou, “Inter-sentence relation extraction with document-level graph convolutional neural network,” 2019. [Online]. Available: https: //arxiv.org/abs/1906.04684

    work page Pith review arXiv 2019

Show all 42 references
  1. [8]

    Document-level relation extraction with adaptive thresholding and localized context pooling,

    W. Zhou, K. Huang, T. Ma, and J. Huang, “Document-level relation extraction with adaptive thresholding and localized context pooling,” Proceedings of the AAAI Conference on Artificial Intelligence , vol. 35, no. 16, pp. 14 612–14 620, May 2021. [Online]. Available: https://ojs...

    2021

  2. [9]

    Entity structure within and throughout: Modeling mention dependencies for document-level relation extraction,

    B. Xu, Q. Wang, Y. Lyu, Y. Zhu, and Z. Mao, “Entity structure within and throughout: Modeling mention dependencies for document-level relation extraction,” Proceedings of the AAAI Conference on Artificial Intelligence , vol. 35, no. 16, pp. 14 149–14 157, May 2021. [Online]. A...

    2021

  3. [10]

    Relation classification via convolutional deep neural network,

    D. Zeng, K. Liu, S. Lai, G. Zhou, and J. Zhao, “Relation classification via convolutional deep neural network,” in Pro- ceedings of COLING 2014, the 25th international conference on computational linguistics: technical papers, 2014, pp. 2335–2344

    2014

  4. [11]

    Bidirectional recurrent con- volutional neural network for relation classification,

    R. Cai, X. Zhang, and H. Wang, “Bidirectional recurrent con- volutional neural network for relation classification,” in Pro- ceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , 2016, pp. 756–765

    2016

  5. [12]

    Prgc: Potential relation and global correspondence based joint relational triple extraction,

    H. Zheng, R. Wen, X. Chen, Y. Yang, Y. Zhang, Z. Zhang, N. Zhang, B. Qin, M. Xu, and Y. Zheng, “Prgc: Potential relation and global correspondence based joint relational triple extraction,” arXiv preprint arXiv:2106.09895, 2021

    2021 arXiv

  6. [13]

    Docred: A large-scale document-level relation extraction dataset,

    Y. Yao, D. Ye, P. Li, X. Han, Y. Lin, Z. Liu, Z. Liu, L. Huang, J. Zhou, and M. Sun, “Docred: A large-scale document-level relation extraction dataset,” arXiv preprint arXiv:1906.06127 , 2019

    1906 arXiv

  7. [14]

    W. Zhou, K. Huang, T. Ma, and J. Huang, “Document-level relation extraction with adaptive thresholding and localized Fig. 9. Output Relations of the proposed method on same entity pair on different Local context context pooling,” in Proceedings of the AAAI conference on artifi...

    2021

  8. [15]

    Connecting the dots: Document-level neural relation extraction with edge- oriented graphs,

    F. Christopoulou, M. Miwa, and S. Ananiadou, “Connecting the dots: Document-level neural relation extraction with edge- oriented graphs,” arXiv preprint arXiv:1909.00228, 2019

    1909 arXiv

  9. [16]

    Distant supervision for relation extraction beyond the sentence boundary,

    C. Quirk and H. Poon, “Distant supervision for relation extraction beyond the sentence boundary,” arXiv preprint arXiv:1609.04873, 2016

    2016 arXiv

  10. [17]

    Reasoning with latent structure refinement for document-level relation extraction,

    G. Nan, Z. Guo, I. Sekuli´ c, and W. Lu, “Reasoning with latent structure refinement for document-level relation extraction,” arXiv preprint arXiv:2005.06312, 2020

    2005 arXiv

  11. [18]

    Global-to-local neural networks for document-level relation extraction,

    D. Wang, W. Hu, E. Cao, and W. Sun, “Global-to-local neural networks for document-level relation extraction,”arXiv preprint arXiv:2009.10359, 2020

    2009 arXiv

  12. [19]

    Double graph based rea- soning for document-level relation extraction,

    S. Zeng, R. Xu, B. Chang, and L. Li, “Double graph based rea- soning for document-level relation extraction,” arXiv preprint arXiv:2009.13752, 2020

    2009 arXiv

  13. [20]

    Graph enhanced dual attention network for document-level relation extraction,

    B. Li, W. Ye, Z. Sheng, R. Xie, X. Xi, and S. Zhang, “Graph enhanced dual attention network for document-level relation extraction,” in Proceedings of the 28th international conference on computational linguistics, 2020, pp. 1551–1560

    2020

  14. [21]

    Document-level relation extraction with dual-tier het- erogeneous graph,

    Z. Zhang, B. Yu, X. Shu, T. Liu, H. Tang, W. Yubin, and L. Guo, “Document-level relation extraction with dual-tier het- erogeneous graph,” in Proceedings of the 28th International Conference on Computational Linguistics, 2020, pp. 1630–1641

    2020

  15. [22]

    Mrn: A locally and globally mention-based reasoning network for document- level relation extraction,

    J. Li, K. Xu, F. Li, H. Fei, Y. Ren, and D. Ji, “Mrn: A locally and globally mention-based reasoning network for document- level relation extraction,” in Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, 2021, pp. 1359– 1370

    2021

  16. [23]

    Document-level relation extrac- tion with reconstruction,

    W. Xu, K. Chen, and T. Zhao, “Document-level relation extrac- tion with reconstruction,” in Proceedings of the AAAI Confer- ence on Artificial Intelligence, vol. 35, no. 16, 2021, pp. 14 167– 14 175

    2021

  17. [24]

    Entity structure within and throughout: Modeling mention dependencies for document-level relation extraction,

    B. Xu, Q. Wang, Y. Lyu, Y. Zhu, and Z. Mao, “Entity structure within and throughout: Modeling mention dependencies for document-level relation extraction,” in Proceedings of the AAAI conference on artificial intelligence , vol. 35, no. 16, 2021, pp. 14 149–14 157

    2021

  18. [25]

    Learning logic rules for document-level relation extraction,

    D. Ru, C. Sun, J. Feng, L. Qiu, H. Zhou, W. Zhang, Y. Yu, and L. Li, “Learning logic rules for document-level relation extraction,” arXiv preprint arXiv:2111.05407, 2021

    2021 arXiv

  19. [26]

    Modular self-supervision for document-level relation extraction,

    S. Zhang, C. Wong, N. Usuyama, S. Jain, T. Naumann, and H. Poon, “Modular self-supervision for document-level relation extraction,” arXiv preprint arXiv:2109.05362, 2021

    2021 arXiv

  20. [27]

    Revis- iting docred–addressing the false negative problem in relation extraction,

    Q. Tan, L. Xu, L. Bing, H. T. Ng, and S. M. Aljunied, “Revis- iting docred–addressing the false negative problem in relation extraction,” arXiv preprint arXiv:2205.12696, 2022

    2022 arXiv

  21. [29]

    Bart: De- noising sequence-to-sequence pre-training for natural language generation, translation, and comprehension,

    M. Lewis, Y. Liu, N. Goyal, M. Ghazvininejad, A. Mo- hamed, O. Levy, V. Stoyanov, and L. Zettlemoyer, “Bart: De- noising sequence-to-sequence pre-training for natural language generation, translation, and comprehension,” arXiv preprint arXiv:1910.13461, 2019

    1910 arXiv

  22. [30]

    Sire: Separate intra-and inter- sentential reasoning for document-level relation extraction,

    S. Zeng, Y. Wu, and B. Chang, “Sire: Separate intra-and inter- sentential reasoning for document-level relation extraction,” arXiv preprint arXiv:2106.01709, 2021

    2021 arXiv

  23. [31]

    Discriminative reason- ing for document-level relation extraction,

    W. Xu, K. Chen, and T. Zhao, “Discriminative reason- ing for document-level relation extraction,” arXiv preprint arXiv:2106.01562, 2021

    2021 arXiv

  24. [32]

    Document-level relation extraction as se- mantic segmentation,

    N. Zhang, X. Chen, X. Xie, S. Deng, C. Tan, M. Chen, F. Huang, L. Si, and H. Chen, “Document-level relation extraction as se- mantic segmentation,” arXiv preprint arXiv:2106.03618, 2021

    2021 arXiv

  25. [33]

    A unified positive- unlabeled learning framework for document-level relation ex- traction with different levels of labeling,

    Y. Wang, X. Liu, W. Hu, and T. Zhang, “A unified positive- unlabeled learning framework for document-level relation ex- traction with different levels of labeling,” arXiv preprint arXiv:2210.08709, 2022

    2022 arXiv

  26. [34]

    Document-level relation extraction with adaptive focal loss and knowledge distillation,

    Q. Tan, R. He, L. Bing, and H. T. Ng, “Document-level relation extraction with adaptive focal loss and knowledge distillation,” arXiv preprint arXiv:2203.10900, 2022

    2022 arXiv

  27. [35]

    Sais: su- pervising and augmenting intermediate steps for document-level relation extraction,

    Y. Xiao, Z. Zhang, Y. Mao, C. Yang, and J. Han, “Sais: su- pervising and augmenting intermediate steps for document-level relation extraction,” arXiv preprint arXiv:2109.12093, 2021

    2021 arXiv

  28. [36]

    Dreeam: Guiding attention with evidence for improving document-level relation extrac- tion,

    Y. Ma, A. Wang, and N. Okazaki, “Dreeam: Guiding attention with evidence for improving document-level relation extrac- tion,” arXiv preprint arXiv:2302.08675, 2023

    2023 arXiv

  29. [37]

    Eider: Em- powering document-level relation extraction with efficient ev- idence extraction and inference-stage fusion,

    Y. Xie, J. Shen, S. Li, Y. Mao, and J. Han, “Eider: Em- powering document-level relation extraction with efficient ev- idence extraction and inference-stage fusion,” arXiv preprint arXiv:2106.08657, 2021

    2021 arXiv

  30. [38]

    Enhancing document- level relation extraction by entity knowledge injection,

    X. Wang, Z. Wang, W. Sun, and W. Hu, “Enhancing document- level relation extraction by entity knowledge injection,” inInter- national Semantic Web Conference. Springer, 2022, pp. 39–56

    2022

  31. [39]

    Reside: Improving distantly-supervised neural relation extraction using side information,

    S. Vashishth, R. Joshi, S. S. Prayaga, C. Bhattacharyya, and P. Talukdar, “Reside: Improving distantly-supervised neural relation extraction using side information,” arXiv preprint arXiv:1812.04361, 2018

    2018 arXiv

  32. [40]

    Recon: relation extraction using knowledge graph context in a graph neural network,

    A. Bastos, A. Nadgeri, K. Singh, I. O. Mulang, S. Shekarpour, J. Hoffart, and M. Kaul, “Recon: relation extraction using knowledge graph context in a graph neural network,” in Pro- ceedings of the Web Conference 2021, 2021, pp. 1673–1685

    2021

  33. [41]

    Injecting knowledge base information into end-to-end joint entity and relation extraction and coreference resolution,

    S. Verlinden, K. Zaporojets, J. Deleu, T. Demeester, and C. De- velder, “Injecting knowledge base information into end-to-end joint entity and relation extraction and coreference resolution,” arXiv preprint arXiv:2107.02286, 2021

    2021 arXiv

  34. [42]

    Revisit- ing document-level relation extraction with context-guided link prediction,

    M. Jain, R. Mutharaju, R. Kavuluru, and K. Singh, “Revisit- ing document-level relation extraction with context-guided link prediction,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 38, no. 16, 2024, pp. 18 327–18 335

    2024

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.