Deterministic pipeline for structuring text into JSON and XML with formal constraints, controlled generation and operational integration

N.N. Oltyan

doi:10.18127/j19997493-202605-13

Deterministic pipeline for structuring text into JSON and XML with formal constraints, controlled generation and operational integration

research article UDC 004.8 Pages: 97-104 doi.org/10.18127/j19997493-202605-13

¹ Financial University under the Government of the Russian Federation (Moscow, Russia)

Keywords: Large language models software architecture deterministic pipeline JSON and XML schemas structured output schema induction reproducibility validation typing

Abstract

The paper presents a software implementation of a deterministic pipeline for structuring text data into JSON and XML. The focus is on data-flow management, formalization of intermediate representations, configuration-based reproducibility and deterministic post-processing procedures.

Goal – describe the implemented software system, including data flow, formal artifacts, correctness control mechanisms and operational integration features.

The pipeline is implemented as a sequence of strictly defined stages with explicit contracts between them. The system includes validation, repair, canonicalization, strict typing and execution tracing mechanisms.

The proposed approach is suitable for applied information systems that require traceable, reproducible and formally checked conversion of unstructured text into structured data.

Received: 09.09.2026
Approved after review: 23.09.2026
Accepted: 01.10.2026

How to cite

Олтян Н.Н. Детерминированный конвейер структурирования текста в JSON и XML с формальными ограничениями, управляемой генерацией и эксплуатационной интеграцией // Динамика сложных систем - XXI век. 2026. Т. 20. № 5. С. 97−104. DOI: https://doi.org/10.18127/j19997493-202605-13

References

Oltyan N.N. Architecture and Algorithm of a Deterministic Pipeline for Structuring JSON and XML Data Using LLMs // Neurocomputers: Development, Application. 2026. V. 28. No. 4. P. 47–53. https://doi.org/10.18127/j19998554-202604-04 (in Russian)
Chen J., Zaharia M., Zou J. FrugalGPT: How to Use Large Language Models While Reducing Cost and Improving Performance. Proceedings of the 2023 Conference on Neural Information Processing Systems (NeurIPS). 2023.
Dagdelen J., Dunn A., Lee S. et al. Structured Information Extraction from Scientific Text with Large Language Models. Nature Communications. 2024. V. 15. Art. 1418.
Fu Y., Peng H., Sabharwal A., Clark P., Khot T. Specialized Reasoning: The Role of Domain-Specific Constraints in LLM Systems. arXiv preprint arXiv:2308.06361. 2023.
Liu Y., Li D., Wang K. et al. Are LLMs Good at Structured Outputs? A Benchmark for Evaluating Structured Output Capabilities in LLMs. Information Processing & Management. 2024. V. 61. No. 5. Art. 103809.
Patel R., Ahmad W., Chen D. Constraint-Aware Decoding for Structured Text Generation. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP). 2023.
Zhang T., Tham I., Hou Z. et al. Human-in-the-loop Schema Induction. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics: System Demonstrations. 2023.
Zhou Y., Wang X., Chen Z. et al. Large Language Models as Data Preprocessors: A Framework for Structured Data Pipelines. arXiv preprint arXiv:2401.01234. 2024.

[1] Oltyan N.N. Architecture and Algorithm of a Deterministic Pipeline for Structuring JSON and XML Data Using LLMs // Neurocomputers: Development, Application. 2026. V. 28. No. 4. P. 47–53. https://doi.org/10.18127/j19998554-202604-04 (in Russian)

[2] Chen J., Zaharia M., Zou J. FrugalGPT: How to Use Large Language Models While Reducing Cost and Improving Performance. Proceedings of the 2023 Conference on Neural Information Processing Systems (NeurIPS). 2023.

[3] Dagdelen J., Dunn A., Lee S. et al. Structured Information Extraction from Scientific Text with Large Language Models. Nature Communications. 2024. V. 15. Art. 1418.

[4] Fu Y., Peng H., Sabharwal A., Clark P., Khot T. Specialized Reasoning: The Role of Domain-Specific Constraints in LLM Systems. arXiv preprint arXiv:2308.06361. 2023.

[5] Liu Y., Li D., Wang K. et al. Are LLMs Good at Structured Outputs? A Benchmark for Evaluating Structured Output Capabilities in LLMs. Information Processing & Management. 2024. V. 61. No. 5. Art. 103809.

[6] Patel R., Ahmad W., Chen D. Constraint-Aware Decoding for Structured Text Generation. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP). 2023.

[7] Zhang T., Tham I., Hou Z. et al. Human-in-the-loop Schema Induction. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics: System Demonstrations. 2023.

[8] Zhou Y., Wang X., Chen Z. et al. Large Language Models as Data Preprocessors: A Framework for Structured Data Pipelines. arXiv preprint arXiv:2401.01234. 2024.

Abstract

How to cite

Related articles