CVFive-system extraction study

Artifacts and provenance

Sources, downloads, and reproducibility record

Download the publication formats and inspect the benchmark, implementation, and artifact provenance used by the five-system analysis.

Reproducibility package

Validated report and data artifacts

Academic report · PDF

73-page fixed-layout publication.

Academic report · ODT

Editable OpenDocument version.

Academic report · DOCX

Word-compatible package with native equations, full tables, and publication figures.

Academic report · ZIP

The same DOCX in a ZIP wrapper to prevent browser or file-type misidentification.

Statistical appendix · XLSX

Five-system evidence tiers and category/segment data.

Result provenance · JSON

Machine-readable traceability checks and removed unsupported claims.

Source lineage

Supplied benchmark and implementation sources

SourcePathSHA-256 / status
Original statistical appendixC:\edocument.repo\benchmarkingmodels\benchextractions.chatgpt\-REPORT-005\ChatGPT_Gemma_Qwen_Academic_Statistical_Appendix.xlsx767156cf4d4e28c4f97cdfba472d21a778ca7107a06949bb09621b5816d9583c
Original expanded Word reportC:\edocument.repo\benchmarkingmodels\benchextractions.chatgpt\-REPORT-005\ChatGPT_Gemma_Qwen_Comparative_Academic_Report_Expanded.docx8e0505b2fa9c74d880064079e0883363a3d89f666eaa2928857479b27865560d
Three-system comparative auditC:\edocument.repo\benchmarkingmodels\benchextractions.chatgpt\-REPORT-005\ChatGPT_Gemma_Qwen_24_Category_Comparative_Audit.xlsx4c372c6f8d6b4b9dc4bc3d2cd7b4731ad17e0b08215641f70eb053083f55d34d
BERT combined workbookC:\edocument.repo\benchmarkingmodels\benchextractions.bert\-REPORT-001\BERT_CV_Extraction_24_Category_Combined_Report.xlsx5c44a1d1cf56631129eecf8420e7c3205387294990c24f3fa3ff73b86fb74c49
BERT academic reportC:\edocument.repo\benchmarkingmodels\benchextractions.bert\-REPORT-001\BERT_CV_Extraction_24_Category_Academic_Report.docx7904cb0420d1ded2ef9af0c36bd46536236aed068d3cab5cd3e571a226460c65
HNLP combined workbookC:\edocument.repo\benchmarkingmodels\benchextractions.hnlp\-REPORT-001\HNLP_CV_Extraction_24_Category_Combined_Report.xlsx273e03301fa746992360276902eb941a03837ae5eba7c5d710a3d223ad9dfa28
HNLP academic reportC:\edocument.repo\benchmarkingmodels\benchextractions.hnlp\-REPORT-001\HNLP_CV_Extraction_24_Category_Academic_Report.docx50b061f9a7a4c49b41b96c62dcac12b50899e1486000187b95bcc6d0c4f44cc9
Hybrid strategyC:\edocument.repo\edoc-260\modules\seniorconsultant-core\src\main\java\org\seniorconsultant\core\extract\HNlpExtractionStrategy.java2251d8be4b7a15c533b9d73c5dbfbd41c9c43f98cd5be7f091156059fd54fd58

The Word report contains the complete seven-file Java source manifest for HNLP and BERT. Exact ONNX/tokenizer identity, optional OpenNLP activation, runtime provider, and per-document configuration remain unresolved unless recorded separately.

Selected literature

Methodological references

  1. Devlin et al. (2019). BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. NAACL-HLT.
  2. Zaratiana et al. (2024). GLiNER: Generalist Model for Named Entity Recognition using Bidirectional Transformer. NAACL.
  3. Apache OpenNLP documentation.
  4. Microsoft ONNX Runtime documentation.
  5. Deep Java Library documentation.

Citation

Cite the study

Critical Comparative Evaluation of Five CV Information-Extraction Pipelines: ChatGPT, Gemma, Qwen, Hybrid NLP, and BERT MULTI. Technical academic report, 2026.

Figure