pith. machine review for the scientific record. sign in

arxiv: 2605.08997 · v1 · submitted 2026-05-09 · 📡 eess.SP

Recognition: 2 theorem links

· Lean Theorem

SEM-RAG: Structure-Preserving Multimodal Graph Compilation and Entropy-Guided Retrieval for Telecommunication Standards

Yuzhi Yang , Lina Bariah , Yuhuan Lu , Hang Zou , M\'erouane Debbah
Authors on Pith no claims yet

Pith reviewed 2026-05-12 02:14 UTC · model grok-4.3

classification 📡 eess.SP
keywords retrievalgraphsem-ragaccuracycompilationembeddedformulasheaders
0
0 comments X

The pith

SEM-RAG compiles telecommunication standards into structure-preserving graphs and uses entropy-guided retrieval to reach 94.1% accuracy on TeleQnA and 93.8% on ORAN-Bench-13K while reducing indexing token usage compared to standard GraphRAG.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Telecom standards contain many tables, conditions, and formulas whose meaning depends on how cells link to headers or how symbols are defined nearby. Turning these documents into plain text for AI retrieval often breaks those connections and leads to incorrect answers. SEM-RAG first runs a layout-aware compiler that builds a graph: each table cell is tied to its row header, column header, any predicates, and its original location, while formulas become operator graphs linked to nearby symbol definitions. This creates typed graph primitives that keep the original dependencies intact. The graph is then compressed with Structural Entropy Minimization, a step that shrinks the structure without using large language models for bottom-up clustering at indexing time. A Jensen-Shannon alignment layer matches incoming queries to the right subgraphs, and a lightweight query controller keeps online costs stable. Experiments on TeleQnA, TSpec-LLM, SPEC5G, and ORAN-Bench-13K report stronger results on table-heavy and formula-heavy questions, with 94.1% accuracy on TeleQnA and 93.8% on ORAN-Bench-13K, plus a large drop in indexing-time token usage versus standard GraphRAG. The work concludes that structure-preserving compilation is a practical requirement for this domain.

Core claim

Experiments on TeleQnA, TSpec-LLM, SPEC5G, and ORAN-Bench-13K show that SEM-RAG improves performance on table-heavy and formula-heavy questions, reaches 94.1% accuracy on TeleQnA and 93.8% on ORAN-Bench-13K, and cuts indexing-time token usage by a wide margin relative to standard GraphRAG. These results indicate that structure-preserving compilation is a practical requirement for retrieval over telecom specifications, not merely an optional preprocessing step.

Load-bearing premise

The layout-aware compiler accurately converts tables, conditions, and formulas into typed graph primitives that preserve all critical dependencies and relationships without introducing errors or information loss.

Figures

Figures reproduced from arXiv: 2605.08997 by Hang Zou, Lina Bariah, M\'erouane Debbah, Yuhuan Lu, Yuzhi Yang.

Figure 1
Figure 1. Figure 1: Conceptual comparison between a standard text-centric GraphRAG pipeline and SEM-RAG. In the left panel, standard GraphRAG starts from one [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of SEM-RAG. Module 4.1 extracts hierarchical paragraph primitives [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Performance on table-heavy and formula-heavy questions, together [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Indexing-time scalability comparison. SEM shifts hierarchy construc [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 6
Figure 6. Figure 6: Effect of removing core SEM-RAG components on held-out QA [PITH_FULL_IMAGE:figures/full_fig_p012_6.png] view at source ↗
read the original abstract

Telecommunication standards pose a unique challenge for retrieval systems, where accuracy depends on semantic relevance as well as on preserving the structural logic embedded in the documents, including structured relationships embedded in tables, conditions, and formulas. When these elements are flattened into text, critical dependencies are lost, leading to unreliable retrieval. In this paper, we present SEM-RAG, an end-to-end retrieval framework built around two design choices. First, a layout-aware compiler converts text, tables, and formulas into typed graph primitives. Each table cell is linked to its row headers, column headers, predicates, and source coordinates, while each formula is converted into an operator graph tied to nearby symbol definitions. Second, the compiled graph is compressed with Structural Entropy Minimization (SEM), which avoids LLM-based bottom-up clustering during indexing. A Jensen-Shannon alignment layer and a lightweight query controller serve as supporting retrieval components that map user queries to the right subgraphs, while keeping online cost stable. Experiments on TeleQnA, TSpec-LLM, SPEC5G, and ORAN-Bench-13K show that SEM-RAG improves performance on table-heavy and formula-heavy questions, reaches 94.1\% accuracy on TeleQnA and 93.8\% on ORAN-Bench-13K, and cuts indexing-time token usage by a wide margin relative to standard GraphRAG. These results indicate that structure-preserving compilation is a practical requirement for retrieval over telecom specifications, not merely an optional preprocessing step.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper introduces SEM-RAG, an end-to-end retrieval framework for telecommunication standards. It features a layout-aware compiler that converts text, tables, conditions, and formulas into typed graph primitives (linking table cells to headers/predicates/coordinates and formulas to operator graphs), followed by Structural Entropy Minimization (SEM) for graph compression without LLM-based clustering, plus a Jensen-Shannon alignment layer and lightweight query controller. Experiments on TeleQnA, TSpec-LLM, SPEC5G, and ORAN-Bench-13K report 94.1% accuracy on TeleQnA and 93.8% on ORAN-Bench-13K, with gains on table-heavy and formula-heavy questions and reduced indexing-time token usage versus standard GraphRAG, concluding that structure-preserving compilation is a practical requirement rather than optional preprocessing.

Significance. If the central claims hold after verification, the work would be significant for advancing RAG systems in technical domains with dense structured content. It highlights efficiency benefits from entropy-guided compression over LLM clustering and provides concrete evidence that preserving table and formula dependencies improves retrieval accuracy in telecom standards, potentially influencing future graph-based approaches for regulatory and specification documents.

major comments (2)
  1. [Experiments section] The experimental evaluation reports concrete accuracy figures (94.1% on TeleQnA, 93.8% on ORAN-Bench-13K) and efficiency gains but supplies no details on experimental setup, baseline implementations, statistical tests, or controls for confounding factors. This gap directly affects verification of the claimed improvements on table-heavy and formula-heavy questions and the conclusion that structure preservation is required.
  2. [Layout-aware compiler description] The central claim that structure-preserving compilation is a practical requirement rests on the layout-aware compiler correctly converting tables, conditions, and formulas into typed graph primitives while preserving all critical dependencies without errors or information loss. The manuscript provides no implementation details, conversion examples, error rates, or ablation studies on information preservation for this component.
minor comments (1)
  1. The abstract states that indexing-time token usage is cut 'by a wide margin' relative to GraphRAG but does not provide quantitative values or specific comparisons; adding these would improve clarity of the efficiency claims.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback and positive evaluation of the work's significance. We address each major comment below and will revise the manuscript to incorporate the requested details and clarifications.

read point-by-point responses

  1. Referee: [Experiments section] The experimental evaluation reports concrete accuracy figures (94.1% on TeleQnA, 93.8% on ORAN-Bench-13K) and efficiency gains but supplies no details on experimental setup, baseline implementations, statistical tests, or controls for confounding factors. This gap directly affects verification of the claimed improvements on table-heavy and formula-heavy questions and the conclusion that structure preservation is required.

    Authors: We agree that additional experimental details are necessary for full reproducibility and verification. In the revised manuscript, we will expand the Experiments section to include: (1) a complete description of the experimental setup, including hardware, software versions, and dataset preprocessing steps; (2) implementation details for all baselines, particularly how standard GraphRAG was configured and run for fair comparison; (3) statistical tests (e.g., McNemar's test or bootstrap confidence intervals) with p-values for accuracy differences; and (4) explicit controls and breakdowns for confounding factors such as query type (table-heavy vs. formula-heavy) and document complexity. These additions will directly support the claims regarding improvements on structured questions. revision: yes

  2. Referee: [Layout-aware compiler description] The central claim that structure-preserving compilation is a practical requirement rests on the layout-aware compiler correctly converting tables, conditions, and formulas into typed graph primitives while preserving all critical dependencies without errors or information loss. The manuscript provides no implementation details, conversion examples, error rates, or ablation studies on information preservation for this component.

    Authors: We acknowledge that the current manuscript describes the layout-aware compiler at a conceptual level without sufficient low-level details. In the revision, we will add: (1) pseudocode or algorithmic steps for converting tables (linking cells to headers/predicates/coordinates), conditions, and formulas (to operator graphs); (2) concrete conversion examples drawn from telecommunication standards excerpts; (3) any measured error rates or failure modes during compilation; and (4) ablation studies quantifying information preservation, such as retrieval accuracy drops when specific structural links are removed. These changes will provide stronger empirical grounding for the claim that structure preservation is required. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical framework evaluated on external benchmarks

full rationale

The paper describes SEM-RAG as a retrieval framework with a layout-aware compiler converting documents to typed graph primitives and Structural Entropy Minimization for compression, followed by empirical evaluation on independent benchmarks (TeleQnA, TSpec-LLM, SPEC5G, ORAN-Bench-13K). Reported accuracies (94.1% on TeleQnA, 93.8% on ORAN-Bench-13K) and token savings are presented as experimental outcomes, not as quantities derived from fitted parameters or self-referential equations. No derivation chain, mathematical predictions, or load-bearing self-citations are indicated in the abstract or description that would reduce claims to inputs by construction. The framework is self-contained as a proposed system with external validation.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

Review is based solely on the abstract; full paper may contain additional parameters and assumptions. The central claim rests on the premise that graph primitives preserve structure and that entropy minimization retains retrieval-relevant information.

free parameters (2)
  • parameters of the lightweight query controller
    The query controller is described as lightweight but its specific tunable parameters or fitting process are not specified in the abstract.
  • weights or thresholds in Jensen-Shannon alignment layer
    The alignment layer for mapping queries to subgraphs likely involves tunable parameters not detailed here.
axioms (2)
  • domain assumption The layout-aware compiler correctly links table cells to row/column headers, predicates, and source coordinates while converting formulas into operator graphs tied to symbol definitions without loss of critical dependencies.
    This is the foundational premise for claiming structure preservation.
  • domain assumption Structural Entropy Minimization compresses the compiled graph while retaining all information necessary for accurate retrieval on table-heavy and formula-heavy queries.
    Assumed to be effective without the information loss that LLM-based clustering might cause.

pith-pipeline@v0.9.0 · 5591 in / 1674 out tokens · 125086 ms · 2026-05-12T02:14:07.819112+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

36 extracted references · 36 canonical work pages · 3 internal anchors

  1. [1]

    Large Generative AI Models for Telecom: The Next Big Thing?

    L. Bariah, Q. Zhao, H. Zou, Y . Tian, F. Bader, and M. Debbah, “Large Generative AI Models for Telecom: The Next Big Thing?”IEEE Communications Magazine, vol. 62, no. 11, pp. 84–90, 2024

    work page 2024

  2. [2]

    Large language model (llm) for telecommunications: A comprehensive survey on principles, key techniques, and opportunities,

    H. Zhouet al., “Large language model (llm) for telecommunications: A comprehensive survey on principles, key techniques, and opportunities,” IEEE Communications Surveys & Tutorials, 2024

    work page 2024

  3. [3]

    Large Language Models in 6G from Standard to On- device Networks,

    H. Zouet al., “Large Language Models in 6G from Standard to On- device Networks,”Nat. Rev. Electr. Eng., vol. 3, pp. 123–134, 2026

    work page 2026

  4. [4]

    Tspec-llm: An open-source dataset for llm understanding of 3gpp specifications,

    R. Nikbakht, M. Benzaghta, and G. Geraci, “Tspec-llm: An open-source dataset for llm understanding of 3gpp specifications,” in2024 IEEE Globecom Workshops (GC Wkshps). IEEE, 2024, pp. 1–6

    work page 2024

  5. [5]

    Oran-bench-13k: An open source benchmark for assessing llms in open radio access networks,

    P. Gajjar and V . K. Shah, “Oran-bench-13k: An open source benchmark for assessing llms in open radio access networks,” in2025 IEEE 22nd Consumer Communications & Networking Conference (CCNC). IEEE, 2025, pp. 1–4

    work page 2025

  6. [6]

    Teleqna: A benchmark dataset to assess large language models telecommunications knowledge,

    A. Maatouk, F. Ayed, N. Piovesan, A. De Domenico, M. Debbah, and Z.- Q. Luo, “Teleqna: A benchmark dataset to assess large language models telecommunications knowledge,”IEEE Network, vol. 40, pp. 253–260, 2026

    work page 2026

  7. [7]

    Gsma open-telco llm benchmarks,

    G. Foundry, “Gsma open-telco llm benchmarks,” 2025. [Online]. Available: https://www.gsma.com/newsroom/press-release/ gsma-open-telco-llm-benchmarks-launches-to-advance-ai-in-telecoms/

    work page 2025

  8. [8]

    Retrieval-Augmented Generation for Large Language Models: A Survey

    Y . Gaoet al., “Retrieval-augmented generation for large language models: A survey,”arXiv preprint arXiv:2312.10997, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  9. [9]

    From Local to Global: A Graph RAG Approach to Query-Focused Summarization

    D. Edge, H. Trinh, N. Cheng, J. Bradley, A. Chao, A. Mody, S. Truitt, D. Metropolitansky, R. O. Ness, and J. Larson, “From local to global: A graph rag approach to query-focused summarization,”arXiv preprint arXiv:2404.16130, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  10. [10]

    Unifying large language models and knowledge graphs: A roadmap,

    S. Pan, L. Luo, Y . Wang, C. Chen, J. Wang, and X. Wu, “Unifying large language models and knowledge graphs: A roadmap,”IEEE Transactions on Knowledge and Data Engineering, vol. 36, no. 7, pp. 3580–3599, 2024

    work page 2024

  11. [11]

    Under- standing telecom language through large language models,

    L. Bariah, H. Zou, Q. Zhao, M. Belkacem, and M. Debbah, “Under- standing telecom language through large language models,” inIEEE Global Communications Conference, 2023, pp. 6542–6547

    work page 2023

  12. [12]

    TelecomGPT: A Framework to Build Telecom-Specific Large Language Models,

    H. Zou, Q. Zhao, Y . Tian, L. Bariah, F. Bader, T. Lestable, and M. Debbah, “TelecomGPT: A Framework to Build Telecom-Specific Large Language Models,”IEEE Transactions on Machine Learning in Communications and Networking, vol. 3, pp. 948–975, 2025

    work page 2025

  13. [13]

    Telcoai: Advancing 3gpp technical specification search through agentic multi- modal retrieval-augmented generation,

    R. Ghosh, C.-H. Liu, G. Rele, V . S. Ravipati, and H. Aouad, “Telcoai: Advancing 3gpp technical specification search through agentic multi- modal retrieval-augmented generation,” inProceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Li...

    work page 2025

  14. [14]

    Telco-rag: Navigating the challenges of retrieval augmented language models for telecommunications,

    A.-L. Bornea, F. Ayed, A. De Domenico, N. Piovesan, and A. Maatouk, “Telco-rag: Navigating the challenges of retrieval augmented language models for telecommunications,” inGLOBECOM 2024-2024 IEEE Global Communications Conference. IEEE, 2024, pp. 2359–2364

    work page 2024

  15. [15]

    Large language models for wireless networks: An overview from the prompt engineering perspective,

    H. Zhouet al., “Large language models for wireless networks: An overview from the prompt engineering perspective,”IEEE Wireless Communications, vol. 32, no. 4, pp. 98–106, 2025

    work page 2025

  16. [16]

    Curricu- lum engineering: Structured learning for large language models (llms) through curriculum based retrieval,

    K. Sun, Z. Zhao, H. Yang, J. Zhang, and G. Q. Huang, “Curricu- lum engineering: Structured learning for large language models (llms) through curriculum based retrieval,”IEEE Transactions on Industrial Informatics, vol. 22, 2026

    work page 2026

  17. [17]

    A 2ra-nsmtsllm: Adversarially aligning retrieval-augmented llms for nonstationary multivariate time series forecasting,

    J. Chu, C. Liu, X. Bai, and J. Tan, “A 2ra-nsmtsllm: Adversarially aligning retrieval-augmented llms for nonstationary multivariate time series forecasting,”IEEE Transactions on Industrial Informatics, vol. 22, no. 3, pp. 1805–1816, 2025

    work page 2025

  18. [18]

    A knowledge-augmented multistage reason- ing approach for wind turbine fault cause analysis,

    Y . Hu, P. Wen, and Y . Dai, “A knowledge-augmented multistage reason- ing approach for wind turbine fault cause analysis,”IEEE Transactions on Industrial Informatics, 2026

    work page 2026

  19. [19]

    Can knowledge-graph- based retrieval augmented generation really retrieve what you need?

    J. Yu, Y . Liu, J. Gu, P. Torr, and D. Zhou, “Can knowledge-graph- based retrieval augmented generation really retrieve what you need?” in The Thirty-ninth Annual Conference on Neural Information Processing Systems, 2025

    work page 2025

  20. [20]

    Regraphrag: Re- organizing fragmented knowledge graphs for multi-perspective retrieval- augmented generation,

    S. Kim, S. J. Hwang, J. Kim, J. Park, and Y . S. Choi, “Regraphrag: Re- organizing fragmented knowledge graphs for multi-perspective retrieval- augmented generation,” inFindings of the Association for Computational Linguistics: EMNLP 2025, 2025, pp. 5426–5443

    work page 2025

  21. [21]

    LightRAG: Simple and Fast Retrieval-Augmented Generation

    Z. Guo, L. Xia, Y . Yu, T. Ao, and C. Huang, “Lightrag: Simple and fast retrieval-augmented generation,”arXiv preprint arXiv:2410.05779, vol. 2, no. 3, 2024

    work page internal anchor Pith review arXiv 2024

  22. [22]

    Pathrag: Pruning graph-based retrieval augmented generation with relational paths,

    B. Chen, Z. Guo, Z. Yang, Y . Chen, J. Chen, Z. Liu, C. Shi, and C. Yang, “Pathrag: Pruning graph-based retrieval augmented generation with relational paths,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 40, no. 36, 2026, pp. 30 183–30 191

    work page 2026

  23. [23]

    Hypergraphrag: Retrieval-augmented generation via hypergraph-structured knowledge representation,

    H. Luoet al., “Hypergraphrag: Retrieval-augmented generation via hypergraph-structured knowledge representation,” inThe Thirty-ninth Annual Conference on Neural Information Processing Systems, 2025

    work page 2025

  24. [24]

    When to use graphs in rag: A comprehensive analysis for graph retrieval-augmented generation,

    Z. Xiang, C. Wu, Q. Zhang, S. Chen, Z. Hong, X. Huang, and J. Su, “When to use graphs in rag: A comprehensive analysis for graph retrieval-augmented generation,”arXiv preprint arXiv:2506.05690, 2025

    work page arXiv 2025

  25. [25]

    Lad-rag: layout-aware dynamic rag for visually-rich document understanding,

    Z. Souratiet al., “Lad-rag: layout-aware dynamic rag for visually-rich document understanding,”arXiv preprint arXiv:2510.07233, 2025

    work page arXiv 2025

  26. [26]

    Superrag: Beyond rag with layout-aware graph modeling,

    C. Yang, D.-K. Vu, M.-T. Nguyen, X.-Q. Nguyen, L. Nguyen, and H. Le, “Superrag: Beyond rag with layout-aware graph modeling,” in Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 3: Industry Track), 2025, pp. 544–557

    work page 2025

  27. [27]

    Infinity parser: Layout aware reinforcement learning for scanned document parsing.arXiv preprint arXiv:2506.03197, 2025

    B. Wanget al., “Infinity parser: Layout aware reinforcement learning for scanned document parsing,”arXiv preprint arXiv:2506.03197, 2025

    work page arXiv 2025

  28. [28]

    Multimodal retrieval- augmented generation: Unified information processing across text, im- age, table, and video modalities,

    N. Drushchak, N. Polyakovska, M. Bautina, T. Semenchenko, J. Ko- scielecki, W. Sykala, and M. Wegrzynowski, “Multimodal retrieval- augmented generation: Unified information processing across text, im- age, table, and video modalities,” inProceedings of the 1st Workshop on Multimodal Augmented Generation via Multimodal Retrieval (MAGMaR 2025), 2025, pp. 59–64

    work page 2025

  29. [29]

    Math formula graph retrieval using con- trastive learning over visual and semantic embeddings,

    B. Amador and R. Zanibbi, “Math formula graph retrieval using con- trastive learning over visual and semantic embeddings,” inProceedings of the 2025 International ACM SIGIR Conference on Innovative Con- cepts and Theories in Information Retrieval (ICTIR), 2025, pp. 230–237

    work page 2025

  30. [30]

    Enhancing technical question answering quality through multimodal document segmentation,

    D. Lvov, I. Smirnov, V . V olokha, A. Laushkina, and A. Boukhanovsky, “Enhancing technical question answering quality through multimodal document segmentation,”IEEE Access, 2026

    work page 2026

  31. [31]

    Structure entropy and resistor graphs,

    A. Li and Y . Pan, “Structure entropy and resistor graphs,”arXiv preprint arXiv:1801.03404, 2018

    work page arXiv 2018

  32. [32]

    Hierarchical and incremental structural entropy minimization for unsupervised social event detection,

    Y . Cao, H. Peng, Z. Yu, and P. S. Yu, “Hierarchical and incremental structural entropy minimization for unsupervised social event detection,” inProceedings of the AAAI conference on artificial intelligence, vol. 38, no. 8, 2024, pp. 8255–8264

    work page 2024

  33. [33]

    Unsuper- vised timeline summarization via time-aware graph structural entropy minimization,

    F. Peng, Y . Xian, H. Wang, Y . Huang, R. Song, and Z. Yu, “Unsuper- vised timeline summarization via time-aware graph structural entropy minimization,” in2025 IEEE International Conference on Big Data (BigData). IEEE, 2025, pp. 589–598

    work page 2025

  34. [34]

    Hyperbolic continuous structural entropy for hierarchical clustering,

    G. Zeng, H. Peng, A. Li, L. Sun, C. Liu, S. Li, Y . Pan, and P. S. Yu, “Hyperbolic continuous structural entropy for hierarchical clustering,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 40, no. 33, 2026, pp. 28 094–28 102

    work page 2026

  35. [35]

    T-retriever: Tree- based hierarchical retrieval augmented generation for textual graphs,

    C. Wei, H. Qin, S. He, Y . Wang, and Y . Chen, “T-retriever: Tree- based hierarchical retrieval augmented generation for textual graphs,” inProceedings of the AAAI Conference on Artificial Intelligence, 2026

    work page 2026

  36. [36]

    Spec5g: A dataset for 5g cellular network protocol analysis,

    I. Karim, K. S. Mubasshir, M. M. Rahman, and E. Bertino, “Spec5g: A dataset for 5g cellular network protocol analysis,” inFindings of the Association for Computational Linguistics: IJCNLP-AACL 2023 (Findings), 2023, pp. 20–38

    work page 2023