Pith. sign in

REVIEW 4 major objections 3 minor 91 references

Extracting Structured Requirements from Unstructured Building Technical Specifications for Building Information Modeling

T0 review · 4 major / 3 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read Pretrained French-language models can extract named entities from building specifications with F1 above 90%, and Random Forest extracts relations above 80%.

desk verdict A plausible applied NLP benchmark for French building specs, but the reported F1 numbers are unverifiable without the corpus details; worth peer review only if the data and annotation protocol are disclosed. read the letter →

arxiv 2508.13833 v1 pith:BJ6RHKS3 submitted 2025-08-19 cs.CL cs.AI

Insaf Nahri , Romain Pinquié , Philippe Véron , Nicolas Bus , Mathieu Thorel This is my paper
classification cs.CLcs.AI
keywords BuildingInformationModelingNamedEntityRecognitionRelationExtractionFrenchTechnicalSpecificationsCamemBERTRandomForestTransferlearningNaturallanguageprocessing
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 tries to show that standard natural-language-processing tools, without domain-specific pretraining, can turn unstructured French Building Technical Specifications into structured BIM-ready data. It tests two named-entity-recognition approaches—CamemBERT and the French language model Fr_core_news_lg—against rule-based and deep-learning baselines, and tests four relation-extraction models including Random Forest. On a hand-crafted annotated dataset, both NER models pass 90% F1 and Random Forest passes 80% F1. If the result holds beyond this dataset, construction firms could automatically extract requirements and feed them into building information models for automated compliance checking.

What carries the argument

Two pretrained French-language encoders carry the NER stage: CamemBERT, a transformer-based language model, and Fr_core_news_lg, a French statistical model, both used with transfer learning to tag requirement entities. The relation-extraction stage uses a custom feature vector fed into four supervised classifiers, with Random Forest the best. The annotated dataset of French Building Technical Specifications supplies the ground-truth labels that the whole comparison depends on.

What would settle it

Collect a fresh sample of French Building Technical Specifications from several different firms, annotate it with the same entity and relation categories, run the published CamemBERT/Fr_core_news_lg and Random Forest pipelines, and compare F1 scores; a drop well below 90% for NER or 80% for RE would falsify the claim as a general result.

Watch

Extended reading notes

Core claim

The central claim is that a two-stage NLP pipeline works on French BTS documents: first tag requirement-related entities (such as materials, quantities, locations, performance values), then classify relations between them. CamemBERT, a French transformer, and Fr_core_news_lg, a French transfer-learning model, are the strongest NER performers, both exceeding 90% F1. For relation extraction, Random Forest trained on a custom feature vector outperforms the other supervised approaches with F1 above 80%. The authors build and evaluate on a hand-crafted annotated corpus of French building specifications, and they frame the extracted output as a step toward a knowledge graph for automatic verificat

Load-bearing premise

The hand-crafted annotated dataset is representative and correctly labeled; the reported F1 scores stand or fall with the quality and realism of those annotations.

Editorial extensions

If this is right

  • A domain-agnostic pretrained model can do requirement extraction in French construction text without retraining on building-specific corpora.
  • The NER+RE pipeline converts free-text clauses into structured requirement records suitable for a knowledge graph.
  • Random Forest's 80% relation F1 suggests lightweight supervised models remain competitive for relation extraction when features are well chosen.
  • The planned knowledge-graph representation would let future systems verify compliance automatically against BIM data.

Reading between the lines

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

  • Because the evaluation uses one hand-crafted dataset, the 90%/80% figures are likely optimistic for heterogeneous real-world BTS documents; a cross-firm test set would reveal the gap.
  • The relation-extraction feature vector could be combined with the NER models' embeddings instead of hand-crafted features, potentially improving recall on long-distance relations.
  • The same transfer-learning trick may transfer to other French technical domains, such as electrical or plumbing specifications, since both NER models were pretrained on general French text.
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

4 major / 3 minor

Summary. The paper addresses the automatic extraction of structured requirements from French Building Technical Specifications (BTS) for use in Building Information Modeling (BIM). It compares several Named Entity Recognition (NER) approaches—including the transformer-based CamemBERT, the spaCy French pipeline Fr_core_news_lg, rule-based methods, and other deep learning baselines—and several Relation Extraction (RE) models, including Random Forest, on a hand-crafted annotated dataset. The abstract reports NER F1-scores above 90% for CamemBERT and Fr_core_news_lg, and an RE F1-score above 80% for Random Forest, with the stated intent of eventually exporting the outputs to a knowledge graph.

Significance. If the reported performance were rigorously substantiated, the paper would provide a useful practical benchmark for French-domain construction NLP, showing that general-domain French language models can be adapted to BIM-related NER and that a relatively simple feature-based RE classifier can achieve competitive accuracy. The comparison of multiple NER and RE approaches on a bespoke corpus is a legitimate starting point for this application area. However, the significance is strongly conditional on the dataset being representative, correctly annotated, and evaluated without leakage. The current manuscript provides only the abstract-level claims; it does not make available the corpus, the annotation protocol, or the evaluation harness, so the external validity of the numbers cannot be assessed. The paper's strength is its explicit use of a hand-crafted dataset and its multi-model comparison; its weakness is that the evidentiary basis for the headline numbers is not presented.

major comments (4)
  1. [Abstract / Full text] The central empirical claim—NER F1 > 90% and RE F1 > 80%—is reported without any dataset statistics. The abstract does not state the number of documents, sentences, tokens, entity types, relation types, or label distributions. The full text, even where readable, does not provide these corpus statistics. Without this information, a 90% F1 score is uninterpretable: a small corpus with few highly regular entity types can yield deceptively high scores. This is a load-bearing gap because the generalization claim about extracting requirements from real-world BTS documents rests on the representativeness of this dataset.
  2. [Evaluation methodology / Full text] The manuscript does not describe the train/dev/test split, cross-validation protocol, or any leakage-prevention measures. It is not stated whether the Random Forest feature vector was designed after inspecting the same annotated corpus, nor whether the annotators who labeled the data also developed the rule-based baselines. The full text contains no legible description of the evaluation procedure, no error bars, and no confidence intervals. As reported, the results are an internal benchmark; they cannot support a claim of superior performance without a clearly defined and leakage-free evaluation setup.
  3. [Data and code availability / Full text] No data or code are provided. The dataset is described only as 'hand-crafted,' with no annotation guidelines, no inter-annotator agreement, and no sample annotation. In an empirical NLP paper, the gold standard is a public dataset or a detailed appendix with annotation rules, label definitions, and examples. Without this, the reported F1 scores are not reproducible and cannot be independently verified. This is a major limitation, not merely a presentation issue.
  4. [Full text legibility] The provided full text is largely corrupted (mojibake), making most sections unreadable. As a result, even method details, hyperparameters, result tables, and any existing discussion of limitations cannot be checked. The reviewing rule requires treating the manuscript as submitted; in its current form, the body of the paper does not support technical scrutiny. The authors must provide a clean, readable version in any revision.
minor comments (3)
  1. [Abstract / Terminology] Fr_core_news_lg is a spaCy processing pipeline, not a transformer model; the phrase 'transfer learning' should be used more precisely, since it is not the same kind of transfer as fine-tuning CamemBERT.
  2. [RE comparison] The abstract states Random Forest 'proved most effective' for RE, but no comparison with the other three supervised RE models is shown in the abstract or legible text. Please report the F1 of all compared models and their hyperparameters.
  3. [Statistical reporting] Report variance across runs or folds, and consider significance testing for the difference between CamemBERT and Fr_core_news_lg. Without such measures, 'superior performance' is informal.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the reported F1 scores are empirical benchmark measurements on an annotated dataset, not predictions derived from fitted quantities or from self-citations.

full rationale

The paper's central claims are empirical measurements: CamemBERT and Fr_core_news_lg achieve NER F1-scores over 90% and Random Forest achieves RE F1 above 80% on a hand-crafted annotated dataset of French Building Technical Specifications. These are evaluation results on a dataset, not quantities derived by construction from the inputs. No equation in the available text defines a target metric in terms of the fitted parameters, no fitted parameter is renamed as a prediction, and no load-bearing argument depends on a self-citation. The abstract explicitly frames the figures as benchmark comparisons ('A hand-crafted annotated dataset is used to compare the effectiveness of NER approaches and RE models'), and the reported F1 scores are measurements of held-out performance. Even though the full text is garbled and the dataset is not publicly described in detail, that is an evidentiary/reproducibility limitation, not circularity. A concern that the hand-crafted dataset may be small or unrepresentative concerns generalization and external validity, not circular reasoning. Therefore the circularity score is 0.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

This is an empirical benchmark paper, so the main epistemic debt lies in the dataset and the evaluation, not in new theoretical entities. No new physical or abstract objects are introduced. The free parameters are the model hyperparameters and the annotation scheme, which are standard in applied ML research.

free parameters (4)
  • CamemBERT hyperparameters
    Learning rate, number of epochs, batch size, and any other training settings for the transformer model. These are typically tuned on a validation set and the chosen values are not given in the abstract.
  • Fr_core_news_lg hyperparameters
    The spaCy French pipeline's configuration, including any fine-tuning or transfer-learning settings.
  • Random Forest feature vector
    The custom feature vector for relation extraction. The specific features and their weights are not described in the abstract.
  • Entity and relation type definitions
    The set of entity categories (e.g., material, quantity) and relation categories (e.g., has_property) chosen by the authors. These determine what the models are trained to detect.
assumptions (3)
  • domain assumption The hand-crafted annotated dataset is ground truth and correctly labeled.
    The results are computed against this dataset, so any annotation errors directly affect the F1 scores. See abstract: 'A hand-crafted annotated dataset is used to compare the effectiveness of NER approaches and RE models.'
  • domain assumption The pre-trained French models transfer well from the general domain to building specifications.
    The paper relies on CamemBERT and Fr_core_news_lg, both pre-trained on general French text, without additional domain-specific pre-training. See abstract: 'pre-trained on a large French corpus in the general domain'.
  • domain assumption The evaluation methodology (train/test split or cross-validation) is appropriate.
    The reported F1 scores require a robust evaluation design to avoid overfitting. The abstract does not describe this design.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Extracting Structured Requirements from Unstructured Building Technical Specifications for Building Information Modeling." pith.science (2026). https://pith.science/paper/BJ6RHKS3

@misc{pith2026250813833,
  author       = {Pith},
  title        = {Pith review of: Extracting Structured Requirements from Unstructured Building Technical Specifications for Building Information Modeling},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BJ6RHKS3}},
  note         = {Machine review of arXiv:2508.13833}
}
read the original abstract

This study explores the integration of Building Information Modeling (BIM) with Natural Language Processing (NLP) to automate the extraction of requirements from unstructured French Building Technical Specification (BTS) documents within the construction industry. Employing Named Entity Recognition (NER) and Relation Extraction (RE) techniques, the study leverages the transformer-based model CamemBERT and applies transfer learning with the French language model Fr\_core\_news\_lg, both pre-trained on a large French corpus in the general domain. To benchmark these models, additional approaches ranging from rule-based to deep learning-based methods are developed. For RE, four different supervised models, including Random Forest, are implemented using a custom feature vector. A hand-crafted annotated dataset is used to compare the effectiveness of NER approaches and RE models. Results indicate that CamemBERT and Fr\_core\_news\_lg exhibited superior performance in NER, achieving F1-scores over 90\%, while Random Forest proved most effective in RE, with an F1 score above 80\%. The outcomes are intended to be represented as a knowledge graph in future work to further enhance automatic verification systems.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

91 extracted references · 72 canonical work pages

  1. [1]

    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 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

  2. [2]

    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 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

  3. [3]

    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 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

  4. [4]

    A deep neural network-based method for deep information extraction using transfer learning strategies to support automated compliance checking

    Ruichuan Zhang and Nora El-Gohary. A deep neural network-based method for deep information extraction using transfer learning strategies to support automated compliance checking. Automation in Construction , 132:103834, 2021

    2021

  5. [5]

    Semantic nlp-based information extraction from construction regulatory documents for automated compliance checking

    Jiansong Zhang and Nora M El-Gohary. Semantic nlp-based information extraction from construction regulatory documents for automated compliance checking. Journal of Computing in Civil Engineering , 30(2):04015014, 2016

    2016

  6. [6]

    Ontology-based automated information extraction from building energy conservation codes

    Peng Zhou and Nora El-Gohary. Ontology-based automated information extraction from building energy conservation codes. Automation in Construction , 74:103--117, 2017

    2017

  7. [7]

    Bidirectional lstm with attention mechanism and convolutional layer for text classification

    Gang Liu and Jiabao Guo. Bidirectional lstm with attention mechanism and convolutional layer for text classification. Neurocomputing , 337:325--338, 2019

    2019

  8. [8]

    Camembert: a tasty french language model

    Louis Martin, Benjamin Muller, Pedro Javier Ortiz Su \'a rez, Yoann Dupont, Laurent Romary, \'E ric Villemonte de La Clergerie, Djam \'e Seddah, and Beno \^ t Sagot. Camembert: a tasty french language model. arXiv preprint arXiv:1911.03894 , 2019

    arXiv 1911

Show all 91 references
  1. [9]

    Review on knowledge extraction from text and scope in agriculture domain

    EA Nismi Mol and MB Santosh Kumar. Review on knowledge extraction from text and scope in agriculture domain. Artificial Intelligence Review , 56(5):4403--4445, 2023

    2023

  2. [10]

    Information extraction from text

    Jing Jiang. Information extraction from text. Mining text data , pages 11--41, 2012

    2012

  3. [11]

    Information extraction and ontology population using car insurance reports

    Hamid Ahaggach, Lylia Abrouk, and Eric Lebon. Information extraction and ontology population using car insurance reports. In International Conference on Information Technology-New Generations , pages 405--411. Springer, 2023

    2023

  4. [12]

    Semantic bim reasoner for the verification of ifc models

    M Fahad, N Bus, and B Fies. Semantic bim reasoner for the verification of ifc models. In eWork and eBusiness in Architecture, Engineering and Construction , pages 361--368. CRC Press, 2018

    2018

  5. [13]

    Information extraction

    Jerry R Hobbs and Ellen Riloff. Information extraction. Handbook of natural language processing , 15:16, 2010

    2010

  6. [14]

    Computational methods for text analysis and text classification

    Hercules Dalianis and Hercules Dalianis. Computational methods for text analysis and text classification. Clinical Text Mining: Secondary Use of Electronic Patient Records , pages 83--96, 2018

    2018

  7. [15]

    A frame semantic overview of nlp-based information extraction for cancer-related ehr notes

    Surabhi Datta, Elmer V Bernstam, and Kirk Roberts. A frame semantic overview of nlp-based information extraction for cancer-related ehr notes. Journal of biomedical informatics , 100:103301, 2019

    2019

  8. [16]

    Automated analysis of free-text comments and dashboard representations in patient experience surveys: a multimethod co-design study

    Carol Rivas, Daria Tkacz, Laurence Antao, Emmanouil Mentzakis, Margaret Gordon, Sydney Anstee, and Richard Giordano. Automated analysis of free-text comments and dashboard representations in patient experience surveys: a multimethod co-design study. 2019

    2019

  9. [17]

    Natural language processing of requirements for model-based product design with enovia/catia v6

    Romain Pinqui \'e , Philippe V \'e ron, Fr \'e d \'e ric Segonds, and Nicolas Crou \'e . Natural language processing of requirements for model-based product design with enovia/catia v6. In Product Lifecycle Management in the Era of Internet of Things: 12th IFIP WG 5.1 Internat...

    2015

  10. [18]

    Extracting medication information from french clinical texts

    Louise Del \'e ger, Cyril Grouin, and Pierre Zweigenbaum. Extracting medication information from french clinical texts. In MEDINFO 2010 , pages 949--953. IOS Press, 2010

    2010

  11. [19]

    A rule-based named entity recognition for drug-related crime news documents

    Khmael Rakm Rahem. A rule-based named entity recognition for drug-related crime news documents . PhD thesis, UKM, Bangi, 2015

    2015

  12. [20]

    Semantic rule-based construction procedural information extraction to guide jobsite sensing and monitoring

    Ran Ren and Jiansong Zhang. Semantic rule-based construction procedural information extraction to guide jobsite sensing and monitoring. Journal of Computing in Civil Engineering , 35(6):04021026, 2021

    2021

  13. [21]

    Rule-based information extraction for mechanical-electrical-plumbing-specific semantic web

    Lang-Tao Wu, Jia-Rui Lin, Shuo Leng, Jiu-Lin Li, and Zhen-Zhong Hu. Rule-based information extraction for mechanical-electrical-plumbing-specific semantic web. Automation in Construction , 135:104108, 2022

    2022

  14. [22]

    Dictionary-based automated information extraction from geological documents using a deep learning algorithm

    Qinjun Qiu, Zhong Xie, Liang Wu, and Liufeng Tao. Dictionary-based automated information extraction from geological documents using a deep learning algorithm. Earth and Space Science , 7(3):e2019EA000993, 2020

    2020

  15. [23]

    Named entity recognition over electronic health records through a combined dictionary-based approach

    Alexandra Pomares Quimbaya, Alejandro Sierra M \'u nera, Rafael Andr \'e s Gonz \'a lez Rivera, Juli \'a n Camilo Daza Rodr \' guez, Oscar Mauricio Mu \ n oz Velandia, Angel Alberto Garcia Pe \ n a, and Cyril Labb \'e . Named entity recognition over electronic health records t...

    2016

  16. [24]

    Biomedical named entity recognition via dictionary-based synonym generalization

    Zihao Fu, Yixuan Su, Zaiqiao Meng, and Nigel Collier. Biomedical named entity recognition via dictionary-based synonym generalization. arXiv preprint arXiv:2305.13066 , 2023

    2023 arXiv

  17. [25]

    Design information extraction from construction specifications to support cost estimation

    Temitope Akanbi and Jiansong Zhang. Design information extraction from construction specifications to support cost estimation. Automation in Construction , 131:103835, 2021

    2021

  18. [26]

    Project-based as-needed information retrieval from unstructured aec documents

    Hongqin Fan, Fan Xue, and Heng Li. Project-based as-needed information retrieval from unstructured aec documents. Journal of Management in Engineering , 31(1):A4014012, 2015

    2015

  19. [27]

    Machine learning for information extraction in informal domains

    Dayne Freitag. Machine learning for information extraction in informal domains. Machine learning , 39:169--202, 2000

    2000

  20. [28]

    A comprehensive study of named entity recognition in chinese clinical text

    Jianbo Lei, Buzhou Tang, Xueqin Lu, Kaihua Gao, Min Jiang, and Hua Xu. A comprehensive study of named entity recognition in chinese clinical text. Journal of the American Medical Informatics Association , 21(5):808--814, 2014

    2014

  21. [29]

    Named entity recognition using hidden markov model (hmm)

    Sudha Morwal, Nusrat Jahan, and Deepti Chopra. Named entity recognition using hidden markov model (hmm). International Journal on Natural Language Computing (IJNLC) Vol , 1, 2012

    2012

  22. [30]

    Automatic recognition of disorders, findings, pharmaceuticals and body structures from clinical text: An annotation and machine learning study

    Maria Skeppstedt, Maria Kvist, Gunnar H Nilsson, and Hercules Dalianis. Automatic recognition of disorders, findings, pharmaceuticals and body structures from clinical text: An annotation and machine learning study. Journal of biomedical informatics , 49:148--158, 2014

    2014

  23. [31]

    Named entity recognition using conditional random fields

    Nita Patil, Ajay Patil, and BV Pawar. Named entity recognition using conditional random fields. Procedia Computer Science , 167:1181--1188, 2020

    2020

  24. [32]

    A proposed model for bengali named entity recognition using maximum entropy markov model incorporated with rich linguistic feature set

    Fahmida Alam and Md Asiful Islam. A proposed model for bengali named entity recognition using maximum entropy markov model incorporated with rich linguistic feature set. In Proceedings of the International Conference on Computing Advancements , pages 1--6, 2020

    2020

  25. [33]

    Medical named entity recognition in arabic text using svm

    Rema Muftah Hamad and Ahmed Mohamed Abushaala. Medical named entity recognition in arabic text using svm. In 2023 IEEE 3rd International Maghreb Meeting of the Conference on Sciences and Techniques of Automatic Control and Computer Engineering (MI-STA) , pages 200--205. IEEE, 2023

    2023

  26. [34]

    Incorporating dictionaries into deep neural networks for the chinese clinical named entity recognition

    Qi Wang, Yangming Zhou, Tong Ruan, Daqi Gao, Yuhang Xia, and Ping He. Incorporating dictionaries into deep neural networks for the chinese clinical named entity recognition. Journal of biomedical informatics , 92:103133, 2019

    2019

  27. [35]

    Character-level neural network for biomedical named entity recognition

    Mourad Gridach. Character-level neural network for biomedical named entity recognition. Journal of biomedical informatics , 70:85--91, 2017

    2017

  28. [36]

    Power entity recognition based on bidirectional long short-term memory and conditional random fields

    Zhixiang Ji, Xiaohui Wang, Changyu Cai, and Hongjian Sun. Power entity recognition based on bidirectional long short-term memory and conditional random fields. Global Energy Interconnection , 3(2):186--192, 2020

    2020

  29. [37]

    Building a large annotated corpus of english: The penn treebank

    Mitch Marcus, Beatrice Santorini, and Mary Ann Marcinkiewicz. Building a large annotated corpus of english: The penn treebank. Computational linguistics , 19(2):313--330, 1993

    1993

  30. [38]

    Introduction to the conll-2003 shared task: Language-independent named entity recognition

    Erik F Sang and Fien De Meulder. Introduction to the conll-2003 shared task: Language-independent named entity recognition. arXiv preprint cs/0306050 , 2003

    2003 arXiv

  31. [39]

    Deep syntax annotation of the sequoia french treebank

    Marie Candito, Guy Perrier, Bruno Guillaume, Corentin Ribeyre, Kar \"e n Fort, Djam \'e Seddah, and \'E ric Villemonte de La Clergerie. Deep syntax annotation of the sequoia french treebank. In International Conference on Language Resources and Evaluation (LREC) , 2014

    2014

  32. [40]

    A survey on deep transfer learning

    Chuanqi Tan, Fuchun Sun, Tao Kong, Wenchang Zhang, Chao Yang, and Chunfang Liu. A survey on deep transfer learning. In Artificial Neural Networks and Machine Learning--ICANN 2018: 27th International Conference on Artificial Neural Networks, Rhodes, Greece, October 4-7, 2018, P...

    2018

  33. [41]

    https://spacy.io/usage/spacy-101, 2023

    spaCy 101: Everything you need to know · spaCy Usage Documentation . https://spacy.io/usage/spacy-101, 2023. Accessed: Sep. 11, 2023

    2023

  34. [42]

    Bert: Pre-training of deep bidirectional transformers for language understanding

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805 , 2018

    2018 arXiv

  35. [43]

    Flaubert: Unsupervised language model pre-training for french

    Hang Le, Lo \" c Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoit Crabb \'e , Laurent Besacier, and Didier Schwab. Flaubert: Unsupervised language model pre-training for french. arXiv preprint arXiv:1912.05372 , 2019

    1912 arXiv

  36. [44]

    Attention is all you need

    Vaswani Ashish. Attention is all you need. arXiv preprint arXiv: 1706.03762 , 2017

    2017 arXiv

  37. [45]

    Pretrained domain-specific language model for general information retrieval tasks in the aec domain

    Zhe Zheng, Xin-Zheng Lu, Ke-Yin Chen, Yu-Cheng Zhou, and Jia-Rui Lin. Pretrained domain-specific language model for general information retrieval tasks in the aec domain. arXiv preprint arXiv:2203.04729 , 2022

    2022 arXiv

  38. [46]

    Improving language understanding by generative pre-training

    Alec Radford, Karthik Narasimhan, Tim Salimans, Ilya Sutskever, et al. Improving language understanding by generative pre-training. 2018

    2018

  39. [47]

    The learnability of in-context learning

    Noam Wies, Yoav Levine, and Amnon Shashua. The learnability of in-context learning. Advances in Neural Information Processing Systems , 36, 2024

    2024

  40. [48]

    2iner: Instructive and in-context learning on few-shot named entity recognition

    Jiasheng Zhang, Xikai Liu, Xinyi Lai, Yan Gao, Shusen Wang, Yao Hu, and Yiqing Lin. 2iner: Instructive and in-context learning on few-shot named entity recognition. In Findings of the Association for Computational Linguistics: EMNLP 2023 , pages 3940--3951, 2023

    2023

  41. [49]

    Relex—relation extraction using dependency parse trees

    Katrin Fundel, Robert K \"u ffner, and Ralf Zimmer. Relex—relation extraction using dependency parse trees. Bioinformatics , 23(3):365--371, 2007

    2007

  42. [50]

    A review of relation extraction

    Nguyen Bach and Sameer Badaskar. A review of relation extraction. Literature review for Language and Statistics II , 2:1--15, 2007

    2007

  43. [51]

    Combining lexical, syntactic, and semantic features with maximum entropy models for extracting relations

    N Kambhatla. Combining lexical, syntactic, and semantic features with maximum entropy models for extracting relations. proceedings of the acl 2004 on interactive poster and demonstration sessions. ACLdemo’04Stroudsburg, PA, USA: Association for Computational Linguistics , 2004

    2004

  44. [52]

    Utd: Classifying semantic relations by combining lexical and semantic resources

    Bryan Rink and Sanda Harabagiu. Utd: Classifying semantic relations by combining lexical and semantic resources. In Proceedings of the 5th international workshop on semantic evaluation , pages 256--259, 2010

    2010

  45. [53]

    Exploring syntactic structured features over parse trees for relation extraction using kernel methods

    Min Zhang, GuoDong Zhou, and Aiti Aw. Exploring syntactic structured features over parse trees for relation extraction using kernel methods. Information processing & management , 44(2):687--701, 2008

    2008

  46. [54]

    Open language learning for information extraction, June 5 2014

    Oren Etzioni, Robert E Bart, Michael D Schmitz, Stephen G Doderland, et al. Open language learning for information extraction, June 5 2014. US Patent App. 14/083,261

    2014

  47. [55]

    Distant supervision for relation extraction without labeled data

    Mike Mintz, Steven Bills, Rion Snow, and Dan Jurafsky. Distant supervision for relation extraction without labeled data. In Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural Language Processing of th...

    2009

  48. [56]

    Distant supervision for relation extraction beyond the sentence boundary

    Chris Quirk and Hoifung Poon. Distant supervision for relation extraction beyond the sentence boundary. arXiv preprint arXiv:1609.04873 , 2016

    2016 arXiv

  49. [57]

    Robust distant supervision relation extraction via deep reinforcement learning

    Pengda Qin, Weiran Xu, and William Yang Wang. Robust distant supervision relation extraction via deep reinforcement learning. arXiv preprint arXiv:1805.09927 , 2018

    2018 arXiv

  50. [58]

    Discovering relations among named entities from large corpora

    Takaaki Hasegawa, Satoshi Sekine, and Ralph Grishman. Discovering relations among named entities from large corpora. In Proceedings of the 42nd Annual Meeting of the Association for Computational Linguistics (acl-04) , pages 415--422, 2004

    2004

  51. [59]

    Unsupervised feature selection for relation extraction

    Jinxiu Chen, Donghong Ji, Chew Lim Tan, and Zheng-Yu Niu. Unsupervised feature selection for relation extraction. In Companion Volume to the Proceedings of Conference including Posters/Demos and tutorial abstracts , 2005

    2005

  52. [60]

    A survey of deep learning methods for relation extraction

    Shantanu Kumar. A survey of deep learning methods for relation extraction. arXiv preprint arXiv:1705.03645 , 2017

    2017 arXiv

  53. [61]

    Multi-path convolutional neural network for distant supervised relation extraction

    Yunyang Li, Zhinong Zhong, and Ning Jing. Multi-path convolutional neural network for distant supervised relation extraction. In Proceedings of the 2nd International Conference on Computer Science and Application Engineering , pages 1--7, 2018

    2018

  54. [62]

    Relation classification via recurrent neural network

    Dongxu Zhang and Dong Wang. Relation classification via recurrent neural network. arXiv preprint arXiv:1508.01006 , 2015

    2015 arXiv

  55. [63]

    Classifying relations via long short term memory networks along shortest dependency paths

    Yan Xu, Lili Mou, Ge Li, Yunchuan Chen, Hao Peng, and Zhi Jin. Classifying relations via long short term memory networks along shortest dependency paths. In Proceedings of the 2015 conference on empirical methods in natural language processing , pages 1785--1794, 2015

    2015

  56. [64]

    Bidirectional long short-term memory networks for relation classification

    Shu Zhang, Dequan Zheng, Xinchen Hu, and Ming Yang. Bidirectional long short-term memory networks for relation classification. In Proceedings of the 29th Pacific Asia conference on language, information and computation , pages 73--78, 2015

    2015

  57. [65]

    Span model for open information extraction on accurate corpus

    Junlang Zhan and Hai Zhao. Span model for open information extraction on accurate corpus. In Proceedings of the AAAI Conference on Artificial Intelligence , volume 34, pages 9523--9530, 2020

    2020

  58. [66]

    Effective modeling of encoder-decoder architecture for joint entity and relation extraction

    Tapas Nayak and Hwee Tou Ng. Effective modeling of encoder-decoder architecture for joint entity and relation extraction. In Proceedings of the AAAI conference on artificial intelligence , volume 34, pages 8528--8535, 2020

    2020

  59. [67]

    Transformer based network for open information extraction

    Jiabao Han and Hongzhi Wang. Transformer based network for open information extraction. Engineering Applications of Artificial Intelligence , 102:104262, 2021

    2021

  60. [68]

    Multi ^2 oie: Multilingual open information extraction based on multi-head attention with bert

    Youngbin Ro, Yukyung Lee, and Pilsung Kang. Multi ^2 oie: Multilingual open information extraction based on multi-head attention with bert. arXiv preprint arXiv:2009.08128 , 2020

    2009 arXiv

  61. [69]

    Negation-based transfer learning for improving biomedical named entity recognition and relation extraction

    Hermenegildo Fabregat, Andres Duque, Juan Martinez-Romo, and Lourdes Araujo. Negation-based transfer learning for improving biomedical named entity recognition and relation extraction. Journal of Biomedical Informatics , 138:104279, 2023

    2023

  62. [70]

    Semantic frame-based information extraction from utility regulatory documents to support compliance checking

    Xin Xu and Hubo Cai. Semantic frame-based information extraction from utility regulatory documents to support compliance checking. In Advances in Informatics and Computing in Civil and Construction Engineering: Proceedings of the 35th CIB W78 2018 Conference: IT in Design, Con...

    2018

  63. [71]

    Model validation using invariant signatures and logic-based inference for automated building code compliance checking

    Jin Wu and Jiansong Zhang. Model validation using invariant signatures and logic-based inference for automated building code compliance checking. Journal of Computing in Civil Engineering , 36(3):04022002, 2022

    2022

  64. [72]

    A semantic approach for automated rule compliance checking in construction industry

    Dongming Guo, Erling Onstein, and Angela Daniela La Rosa. A semantic approach for automated rule compliance checking in construction industry. IEEE Access , 9:129648--129660, 2021

    2021

  65. [73]

    Model validation for automated building code compliance checking

    Jin Wu, Jiansong Zhang, and Luciana Debs. Model validation for automated building code compliance checking. In Construction Research Congress 2022 , pages 640--650, 2022

    2022

  66. [74]

    o nfelder and Markus K \

    Phillip Sch \"o nfelder and Markus K \"o nig. Deep learning-based entity recognition in construction regulatory documents. In ISARC. Proceedings of the International Symposium on Automation and Robotics in Construction , volume 38, pages 387--394. IAARC Publications, 2021

    2021

  67. [75]

    Automated construction specification review with named entity recognition using natural language processing

    Seonghyeon Moon, Gitaek Lee, Seokho Chi, and Hyunchul Oh. Automated construction specification review with named entity recognition using natural language processing. Journal of Construction Engineering and Management , 147(1):04020147, 2021

    2021

  68. [76]

    Bridge damage recognition from inspection reports using ner based on recurrent neural network with active learning

    Seonghyeon Moon, Sehwan Chung, and Seokho Chi. Bridge damage recognition from inspection reports using ner based on recurrent neural network with active learning. Journal of Performance of Constructed Facilities , 34(6):04020119, 2020

    2020

  69. [77]

    Automated system for construction specification review using natural language processing

    Seonghyeon Moon, Gitaek Lee, and Seokho Chi. Automated system for construction specification review using natural language processing. Advanced Engineering Informatics , 51:101495, 2022

    2022

  70. [78]

    Enhancement and validation of ifcowl ontology based on shapes constraint language (shacl)

    Chaoyue Wang, Liang Zhang, and Wei Yan. Enhancement and validation of ifcowl ontology based on shapes constraint language (shacl). Automation in Construction , 160:105293, 2024

    2024

  71. [79]

    The levenshtein algorithm

    Cuelogic Insights . The levenshtein algorithm. https://www.cuelogic.com/blog/the-levenshtein-algorithm, 2023. Accessed: November 6, 2023

    2023

  72. [80]

    https://plan-bim-2022.fr/actions/ptnb-axe-c-le-dictionnaire-national-de-proprietes-pobim/le-dictionnaire-national-de-proprietes-pobim/, 2022

    Plan bim 2022 - le dictionnaire national de propriétés pobim. https://plan-bim-2022.fr/actions/ptnb-axe-c-le-dictionnaire-national-de-proprietes-pobim/le-dictionnaire-national-de-proprietes-pobim/, 2022. Accessed: November 25, 2022

    2022

  73. [81]

    Le cen, pour une norme européenne en matière d’interopérabilité

    Didier Balaguer. Le cen, pour une norme européenne en matière d’interopérabilité. https://www.datbim.com/2017/03/14/le-cen-pour-une-norme-europeenne-en-matiere-dinteroperabilite/, 3 2017. Accessed: September 12, 2023

    2017

  74. [82]

    Rapport Mission Numérique Bâtiment

    Bertrand DELCAMBRE . Rapport Mission Numérique Bâtiment . https://www.actu-environnement.com/media/pdf/news-23398-rapport-mission-numerique-batiment.pdf, Unknown Year. Accessed: Accessed: September 12, 2023]

    2023

  75. [83]

    The impact of using different annotation schemes on named entity recognition

    Nasser Alshammari and Saad Alanazi. The impact of using different annotation schemes on named entity recognition. Egyptian Informatics Journal , 22(3):295--302, 2021

    2021

  76. [84]

    Named entity recognition in estonian

    Alexander Tkachenko, Timo Petmanson, and Sven Laur. Named entity recognition in estonian. In Proceedings of the 4th Biennial International Workshop on Balto-Slavic Natural Language Processing , pages 78--83, 2013

    2013

  77. [85]

    Fine-grained named entity recognition using conditional random fields for question answering

    Changki Lee, Yi-Gyu Hwang, Hyo-Jung Oh, Soojong Lim, Jeong Heo, Chung-Hee Lee, Hyeon-Jin Kim, Ji-Hyun Wang, and Myung-Gil Jang. Fine-grained named entity recognition using conditional random fields for question answering. In Information Retrieval Technology: Third Asia Informa...

    2006

  78. [86]

    Incorporating non-local information into information extraction systems by gibbs sampling

    Jenny Rose Finkel, Trond Grenager, and Christopher D Manning. Incorporating non-local information into information extraction systems by gibbs sampling. In Proceedings of the 43rd annual meeting of the association for computational linguistics (ACL’05) , pages 363--370, 2005

    2005

  79. [87]

    Roberta: A robustly optimized bert pretraining approach

    Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. Roberta: A robustly optimized bert pretraining approach. arXiv preprint arXiv:1907.11692 , 2019

    1907 arXiv

  80. [88]

    Multiple features for clinical relation extraction: A machine learning approach

    Ilseyar Alimova and Elena Tutubalina. Multiple features for clinical relation extraction: A machine learning approach. Journal of biomedical informatics , 103:103382, 2020

    2020

  81. [89]

    word2vec parameter learning explained

    Xin Rong. word2vec parameter learning explained. arXiv preprint arXiv:1411.2738 , 2014

    2014 arXiv

  82. [90]

    Chemical-protein relation extraction with ensembles of svm, cnn, and rnn models

    Yifan Peng, Anthony Rios, Ramakanth Kavuluru, and Zhiyong Lu. Chemical-protein relation extraction with ensembles of svm, cnn, and rnn models. arXiv preprint arXiv:1802.01255 , 2018

    2018 arXiv

  83. [91]

    Biomedical document relation extraction with prompt learning and knn

    Di Zhao, Yumeng Yang, Peng Chen, Jiana Meng, Shichang Sun, Jian Wang, and Hongfei Lin. Biomedical document relation extraction with prompt learning and knn. Journal of Biomedical Informatics , 145:104459, 2023

    2023

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.