arxiv: 2605.13236 · v1 · submitted 2026-05-13 · 💻 cs.CL

Recognition: unknown

A Hybrid Framework for Natural Language Querying of IFC Models with Relational and Graph Representations

Haowen Xu, Johnson Xuesong Shen, Rabindra Lamsal, Sisi Zlatanova, Yafei Sun

Authors on Pith no claims yet

Pith reviewed 2026-05-14 19:10 UTC · model grok-4.3

classification 💻 cs.CL

keywords IFC modelsnatural language queryingBIMlarge language modelsrelational representationgraph representationhybrid framework

0 comments

The pith

Transforming IFC models into relational tables and graphs lets an LLM answer natural language queries about building data with over 93 percent first-attempt accuracy.

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

The paper presents IfcLLM, a framework that converts Industry Foundation Classes building models into a relational form holding element properties and geometry plus a graph form capturing topological links. These two representations are supplied together to an open-weight large language model that repeatedly refines its answers through retry loops until a satisfactory result appears. Tests on three IFC models drawn from thirty standard scenarios produced first-try accuracy between 93.3 and 100 percent, with every remaining error resolved by switching to a fallback model. The design targets routine building information queries so that users without IFC expertise can still extract answers directly from BIM data.

Core claim

IfcLLM transforms IFC models into complementary relational and graph representations and integrates them via iterative retry-and-refine LLM reasoning to support accurate natural language queries on BIM data.

What carries the argument

Hybrid transformation of IFC data into a relational representation for structured properties and geometry together with a graph representation for topology, processed by iterative retry-and-refine LLM reasoning.

If this is right

Routine BIM analysis tasks become answerable through natural language without IFC expertise.
Open-weight models can be used for reproducible deployment in AEC workflows.
All query failures are recoverable through a simple fallback LLM step.
Complementary representations raise first-attempt success rates above 93 percent on standard scenarios.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same hybrid conversion plus iteration pattern could be tested on other structured engineering data such as infrastructure or product models.
If the iterative correction works reliably, the approach might reduce training requirements for non-specialist staff who need occasional access to BIM repositories.
Extending the framework to handle incremental model updates would allow live querying of evolving construction projects.

Load-bearing premise

The conversions to relational and graph forms retain every detail required for correct answers, and the LLM can spot and fix its mistakes iteratively without creating fresh errors.

What would settle it

A query on one of the three test IFC models where critical information is missing from both representations and iteration still yields an incorrect answer.

Figures

Figures reproduced from arXiv: 2605.13236 by Haowen Xu, Johnson Xuesong Shen, Rabindra Lamsal, Sisi Zlatanova, Yafei Sun.

**Figure 2.** Figure 2: Example graph types for encoding topological relationships between building elements. [PITH_FULL_IMAGE:figures/full_fig_p014_2.png] view at source ↗

**Figure 3.** Figure 3: Representative relational tables in IfcLLM, each shown with a subset of records from the [PITH_FULL_IMAGE:figures/full_fig_p020_3.png] view at source ↗

**Figure 4.** Figure 4: A simple 2-storey house (Building B1) with a single stair and multiple doors. The IFC [PITH_FULL_IMAGE:figures/full_fig_p023_4.png] view at source ↗

**Figure 5.** Figure 5: A duplex building (Building B2) with two separate apartments, each having its own [PITH_FULL_IMAGE:figures/full_fig_p024_5.png] view at source ↗

**Figure 6.** Figure 6: An office building (Building B3) with 99 rooms, 102 doors and 4 staircases. The IFC [PITH_FULL_IMAGE:figures/full_fig_p025_6.png] view at source ↗

**Figure 7.** Figure 7: Visualization of the IfcLLM-generated shortest path between Room A101 and Room [PITH_FULL_IMAGE:figures/full_fig_p031_7.png] view at source ↗

**Figure 8.** Figure 8: Comparison of equivalent shortest-path computation queries generated by GPT OSS [PITH_FULL_IMAGE:figures/full_fig_p035_8.png] view at source ↗

read the original abstract

Building Information Modeling (BIM) is widely used in the Architecture, Engineering, and Construction (AEC) industry, but the complexity of Industry Foundation Classes (IFC) limits accessibility for non-expert users. To address this, we introduce IfcLLM, a hybrid framework for natural language interaction with IFC-based BIM models. It transforms IFC models into complementary representations: a relational representation for structured element properties and geometry, and a graph representation for topological relationships. These representations are integrated through iterative retry-and-refine LLM reasoning. We implement the framework using an open-weight LLM (GPT OSS 120B), supporting reproducible and deployment-oriented workflows. Evaluation on three IFC models with queries derived from 30 scenarios shows first-attempt accuracy of 93.3%-100%, with all failures recovered using a fallback LLM. The results show that combining complementary representations with iterative reasoning enables more accessible natural language querying of IFC data while supporting routine BIM analysis tasks.

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.

Desk Editor's Note private letter to a colleague

IfcLLM gives a workable hybrid of relational and graph views plus iterative LLM fixes for natural language IFC queries, but the evaluation stays too thin to judge how far it generalizes.

read the letter

The paper's main new piece is IfcLLM, which converts an IFC model into a relational table for properties and geometry and a graph for topology, then runs natural language questions through an LLM with a retry-and-refine loop. They test on three models drawn from 30 scenarios and report first-try accuracy from 93.3 to 100 percent, with the rest recovered by a fallback model. Using an open-weight LLM keeps the setup reproducible and deployment-friendly. That dual-representation choice plus the explicit iteration step is the concrete addition over earlier single-rep or plain-prompting attempts in the BIM literature. It targets a real pain point: non-experts in AEC who need to pull routine information from complex IFC files without learning the schema. The reported numbers suggest the combination can handle the mix of structured data and relationships at least on the chosen cases. The evaluation is the clear weak point. Three models is a narrow base, and the abstract gives no baselines, no description of how the queries were generated, and no error breakdown. Without those, it is hard to tell how much the hybrid actually improves on simpler LLM access or whether the conversions drop any geometric or topological details that later queries would need. The stress-test worry about information loss during the IFC-to-relational and IFC-to-graph steps is reasonable and needs direct checking in the methods. Citation coverage looks standard for the area and does not raise red flags. This is aimed at researchers and practitioners working on BIM interfaces and applied LLM use in engineering domains. A reader who wants concrete implementation ideas for domain-specific data access would find the framework description useful. I would send it for peer review so referees can press on the evaluation design and representation completeness.

Circularity Check

0 steps flagged

No circularity: empirical framework with external evaluation

full rationale

The paper describes IfcLLM as a hybrid framework that converts IFC models to relational and graph representations then integrates them via iterative LLM reasoning. Evaluation uses three external IFC models and 30 query scenarios, reporting first-attempt accuracies of 93.3-100% with fallback recovery. No equations, fitted parameters, self-referential definitions, or load-bearing self-citations appear in the derivation. The claim that representations are complementary and preserve needed information is an empirical assumption tested on held-out models rather than derived by construction from the inputs. The work is therefore self-contained against external benchmarks with no reduction of predictions to fitted values or prior self-citations.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

Central claim rests on the domain assumption that LLMs can perform reliable iterative reasoning over the dual representations; no free parameters are introduced and the only invented entity is the framework itself.

axioms (1)

domain assumption Large language models can perform reliable iterative retry-and-refine reasoning over structured relational and graph representations derived from IFC models
Invoked to justify the integration step and fallback recovery mechanism described in the abstract.

invented entities (1)

IfcLLM no independent evidence
purpose: Hybrid framework enabling natural language querying of IFC models via dual representations and iterative LLM reasoning
The proposed system itself; no independent falsifiable evidence outside the paper's own evaluation is provided.

pith-pipeline@v0.9.0 · 5478 in / 1293 out tokens · 38824 ms · 2026-05-14T19:10:11.336493+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

88 extracted references · 53 canonical work pages · 5 internal anchors

[1]

C. M. Eastman, BIM handbook: A guide to building information modeling for owners, managers, designers, engineers and contractors, John Wiley & Sons,
[2]

doi:10.1002/9781119287568

work page doi:10.1002/9781119287568
[3]

H. Wang, Y. Pan, X. Luo, Integration of bim and gis in sustainable built environment: A review and bibliometric analysis, Automation in construction 103 (2019) 41–52. doi:10.1016/j.autcon.2019.03.005

work page doi:10.1016/j.autcon.2019.03.005 2019
[4]

Alattas, E

A. Alattas, E. Kalogianni, T. Alzahrani, S. Zlatanova, P. van Oosterom, Map- ping private, common, and exclusive common spaces in buildings from bim/ifc to ladm. a case study from saudi arabia, Land Use Policy 104 (2021) 105355. doi:10.1016/j.landusepol.2021.105355

work page doi:10.1016/j.landusepol.2021.105355 2021
[5]

L. Liu, B. Li, S. Zlatanova, P. van Oosterom, Indoor navigation supported by the industry foundation classes (ifc): A survey, Automation in Construction 121 (2021) 103436. doi:10.1016/j.autcon.2020.103436

work page doi:10.1016/j.autcon.2020.103436 2021
[6]

Aleksandrov, S

M. Aleksandrov, S. Zlatanova, D. J. Heslop, A. Diakite, Bim-based connectivity graph and voxels classification for pedestrian-hazard interaction, Journal of Spatial Science 69 (2024) 699–719. doi:10.1080/14498596.2023.2281923

work page doi:10.1080/14498596.2023.2281923 2024
[7]

H. Xu, F. Omitaomu, S. Sabri, S. Zlatanova, X. Li, Y. Song, Leveraging genera- tive ai for urban digital twins: a scoping review on the autonomous generation of urban data, scenarios, designs, and 3d city models for smart city advancement, Urban Informatics 3 (2024) 29. doi:10.1007/s44212-024-00060-w

work page doi:10.1007/s44212-024-00060-w 2024
[8]

Irizarry, E

J. Irizarry, E. P. Karan, F. Jalaei, Integrating bim and gis to improve the 43 visual monitoring of construction supply chain management, Automation in construction 31 (2013) 241–254. doi:10.1016/j.autcon.2012.12.005

work page doi:10.1016/j.autcon.2012.12.005 2013
[9]

J. B. Kim, W. Jeong, M. J. Clayton, J. S. Haberl, W. Yan, Developing a physical bim library for building thermal energy simulation, Automation in construction 50 (2015) 16–28. doi:10.1016/j.autcon.2014.10.011

work page doi:10.1016/j.autcon.2014.10.011 2015
[10]

J. Won, J. C. Cheng, Identifying potential opportunities of building information modeling for construction and demolition waste management and minimization, Automation in Construction 79 (2017) 3–18. doi:10.1016/j.autcon.2017.02. 002

work page doi:10.1016/j.autcon.2017.02 2017
[11]

A. A. Diakité, S. Zlatanova, Spatial subdivision of complex indoor environments for 3d indoor navigation, International Journal of Geographical Information Science 32 (2018) 213–235. doi:10.1080/13658816.2017.1376066

work page doi:10.1080/13658816.2017.1376066 2018
[12]

doi:10.1016/j.autcon.2019.04.022

A.M.Hasan, A.A.Torky, Y.F.Rashed, Geometricallyaccuratestructuralanal- ysis models in bim-centered software, Automation in construction 104 (2019) 299–321. doi:10.1016/j.autcon.2019.04.022

work page doi:10.1016/j.autcon.2019.04.022 2019
[13]

buildingSMART International, Industry foundation classes (ifc),https:// technical.buildingsmart.org/standards/ifc/, 2025

2025
[14]

Isikdag, S

U. Isikdag, S. Zlatanova, J. Underwood, A bim-oriented model for supporting indoor navigation requirements, Computers, Environment and Urban Systems 41 (2013) 112–123. doi:10.1016/j.compenvurbsys.2013.05.001

work page doi:10.1016/j.compenvurbsys.2013.05.001 2013
[15]

Z. Ma, Z. Liu, Ontology-and freeware-based platform for rapid development of bim applications with reasoning support, Automation in Construction 90 (2018) 1–8. doi:10.1016/j.autcon.2018.02.004. 44

work page doi:10.1016/j.autcon.2018.02.004 2018
[16]

Lin, Z.-Z

J.-R. Lin, Z.-Z. Hu, J.-P. Zhang, F.-Q. Yu, A natural-language-based approach to intelligent data retrieval and representation for cloud bim, Computer-Aided Civil and Infrastructure Engineering 31 (2016) 18–33. doi:10.1111/mice.12151

work page doi:10.1111/mice.12151 2016
[17]

S. Shin, R. R. Issa, Bimasr: Framework for voice-based bim information retrieval, Journal of Construction Engineering and Management 147 (2021) 04021124. doi:10.1061/(ASCE)CO.1943-7862.0002138

work page doi:10.1061/(asce)co.1943-7862.0002138 2021
[18]

M. Yin, L. Tang, C. Webster, S. Xu, X. Li, H. Ying, An ontology-aided, natural language-based approach for multi-constraint bim model querying, Journal of Building Engineering 76 (2023) 107066. doi:10.1016/j.jobe.2023.107066

work page doi:10.1016/j.jobe.2023.107066 2023
[20]

W. X. Zhao, K. Zhou, J. Li, T. Tang, X. Wang, Y. Hou, Y. Min, B. Zhang, J. Zhang, Z. Dong, et al., A survey of large language models, arXiv preprint arXiv:2303.18223 1 (2023). doi:10.48550/arXiv.2303.18223

work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2303.18223 2023
[21]

GPT-4 Technical Report

J. Achiam, S. Adler, S. Agarwal, L. Ahmad, I. Akkaya, F. L. Aleman, D. Almeida, J. Altenschmidt, S. Altman, S. Anadkat, et al., Gpt-4 techni- cal report, arXiv preprint arXiv:2303.08774 (2023). doi:10.48550/arXiv.2303. 08774

work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2303 2023
[22]

URL:https://ai.meta.com/blog/ llama-4-multimodal-intelligence/

Meta, The llama 4 herd: The beginning of a new era of natively multi- modal ai innovation, AI at Meta (2025). URL:https://ai.meta.com/blog/ llama-4-multimodal-intelligence/. 45

2025
[23]

Shi, Y.-S

X. Shi, Y.-S. Liu, G. Gao, M. Gu, H. Li, Ifcdiff: A content-based automatic comparison approach for ifc files, Automation in Construction 86 (2018) 53–68. doi:10.1016/j.autcon.2017.10.013

work page doi:10.1016/j.autcon.2017.10.013 2018
[24]

Iranmanesh, H

S. Iranmanesh, H. Saadany, E. Vakaj, Llm-assisted graph-rag information ex- traction from ifc data, arXiv preprint arXiv:2504.16813 (2025). doi:10.48550/ arXiv.2504.16813

work page arXiv 2025
[25]

Pacheco, C

L. Pacheco, C. Berkmiller, Bimconverse – graphrag for ifc nat- ural language queries, 2024. URL:https://blog.iaac.net/ bimconverse-graphrag-for-ifc-natural-language-queries/

2024
[26]

J. Lin, K. Chen, P. Pan, D. Wu, J. Zhang, Chatbim: Interactive data retrieval andvisualizationfrombim, in: Proceedingsofthe10thInternationalConference on Innovative Production and Construction (IPC 2024), Perth, Australia, 2024, pp. 265–266. doi:10.13140/RG.2.2.31563.20006

work page doi:10.13140/rg.2.2.31563.20006 2024
[27]

Y. Pan, M. Wang, L. Lu, R. Lamsal, E. Pärn, S. Zlatanova, I. Brilakis, Llm- enabled multi-agent framework for natural language interaction with graph- based digital twins, Automation in Construction 183 (2026) 106791. doi:10. 1016/j.autcon.2026.106791

work page arXiv 2026
[28]

Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks

P. Lewis, E. Perez, A. Piktus, F. Petroni, V. Karpukhin, N. Goyal, H. Küttler, M. Lewis, W.-t. Yih, T. Rocktäschel, et al., Retrieval-augmented generation for knowledge-intensive nlp tasks, Advances in neural information processing systems 33 (2020) 9459–9474. doi:10.48550/arXiv.2005.11401

work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2005.11401 2020
[29]

Noardo, K

F. Noardo, K. Arroyo Ohori, T. Krijnen, J. Stoter, An inspection of ifc models from practice, Applied Sciences 11 (2021) 2232. 46

2021
[30]

URL:https://ifcopenshell.org/

IfcOpenShell contributors, IfcOpenShell: The open source IFC toolkit and ge- ometry engine, 2025. URL:https://ifcopenshell.org/

2025
[31]

com/ifcquery/ifcplusplus, 2025

ifcquery contributors, Ifcplusplus: Ifc reader/writer for c++,https://github. com/ifcquery/ifcplusplus, 2025

2025
[32]

xBIM contributors, xBIM Toolkit Documentation,https://docs.xbim.net/, 2025

2025
[33]

Ara 3D contributors, Ara 3D IFC Toolkit,https://github.com/ara3d/ IFC-toolkit, 2025

2025
[34]

ThatOpen contributors, Web IFC,https://github.com/ThatOpen/engine\ _web-ifc, 2025

2025
[35]

ifcopenshell.org/autoapi/ifcpatch/recipes/Ifc2Sql/index.html, 2025

IfcOpenShell contributors, Ifc2Sql Documentation,https://docs. ifcopenshell.org/autoapi/ifcpatch/recipes/Ifc2Sql/index.html, 2025

2025
[36]

Adachi, Outline of ifc model server development project,

Y. Adachi, Outline of ifc model server development project,
[37]

URL:https://www.scribd.com/document/721764681/ Outline-of-Ifc-Model-Server-Development-Project

work page arXiv
[38]

S.-J. You, D. Yang, C. M. Eastman, Relational db implementation of step based product model, in: CIB World Building Congress 2004, Citeseer, 2004, pp. 2–7. URL:https://www.irbnet.de/daten/iconda/CIB9696.pdf

2004
[39]

Solihin, C

W. Solihin, C. Eastman, Y.-C. Lee, D.-H. Yang, A simplified relational database schema for transformation of bim data into a query-efficient and spatially enabled database, Automation in Construction 84 (2017) 367–383. doi:10.1016/j.autcon.2017.10.002. 47

work page doi:10.1016/j.autcon.2017.10.002 2017
[40]

Alzraiee, Cost estimate system using structured query language in bim, In- ternational Journal of Construction Management 22 (2022) 2731–2743

H. Alzraiee, Cost estimate system using structured query language in bim, In- ternational Journal of Construction Management 22 (2022) 2731–2743. doi:10. 1080/15623599.2020.1823061

work page arXiv 2022
[41]

Demirdöğen, Z

G. Demirdöğen, Z. Işık, Y. Arayici, Bim-based big data analytic system for healthcare facility management, Journal of Building Engineering 64 (2023) 105713. doi:10.1016/j.jobe.2022.105713

work page doi:10.1016/j.jobe.2022.105713 2023
[42]

H. Li, H. Liu, Y. Liu, Y. Wang, An object-relational ifc storage model based on oracle database, The International Archives of the Photogram- metry, Remote Sensing and Spatial Information Sciences 41 (2016) 625–631. doi:10.5194/isprs-archives-XLI-B2-625-2016

work page doi:10.5194/isprs-archives-xli-b2-625-2016 2016
[43]

Beetz, L

J. Beetz, L. van Berlo, R. de Laat, P. van den Helm, Bimserver.org–an open source ifc model server, in: Proceedings of the CIP W78 confer- ence, volume 8, 2010. URL:https://research.tue.nl/en/publications/ bimserverorg-an-open-source-ifc-model-server/

2010
[44]

Hijazi, M

I. Hijazi, M. Ehlers, S. Zlatanova, T. Becker, L. van Berlo, Initial investigations for modeling interior utilities within 3d geo context: Transforming ifc-interior utility to citygml/utilitynetworkade, in: Advances in 3D Geo-information sci- ences, Springer, 2010, pp. 95–113. doi:10.1007/978-3-642-12670-3_6

work page doi:10.1007/978-3-642-12670-3_6 2010
[45]

Stouffs, H

R. Stouffs, H. Tauscher, F. Biljecki, Achieving complete and near-lossless con- version from ifc to citygml, ISPRS International Journal of Geo-Information 7 (2018) 355. doi:10.3390/ijgi7090355

work page doi:10.3390/ijgi7090355 2018
[46]

Aggarwal, Y

F.Biljecki, J.Lim, J.Crawford, D.Moraru, H.Tauscher, A.Konde, K.Adouane, S. Lawrence, P. Janssen, R. Stouffs, Extending citygml for ifc-sourced 3d 48 city models, Automation in Construction 121 (2021) 103440. doi:10.1016/j. autcon.2020.103440

work page doi:10.1016/j 2021
[47]

Gröger, L

G. Gröger, L. Plümer, Citygml–interoperable semantic 3d city models, ISPRS Journal of Photogrammetry and Remote Sensing 71 (2012) 12–33. doi:10.1016/ j.isprsjprs.2012.04.004

2012
[48]

A. A. Diakite, L. Díaz-Vilariño, F. Biljecki, Ü. Isikdag, S. Simmons, K. Li, S. Zlatanova, Ifc2indoorgml: An open-source tool for generating indoorgml from ifc, The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences 43 (2022) 295–301

2022
[49]

J. Zhu, M. O. Wong, N. Nisbet, J. Xu, T. Kelly, S. Zlatanova, I. Brilakis, Semantics-basedconnectivitygraphforindoorpathfindingpoweredbyifc-graph, Automation in Construction 171 (2025) 106019. doi:10.1016/j.autcon.2025. 106019

work page doi:10.1016/j.autcon.2025 2025
[50]

J. Zhu, N. Nisbet, M. Yin, R. Wei, I. Brilakis, Releasing the power of graph for building knowledge discovery, arXiv preprint arXiv:2405.16345 (2024). doi:10. 1016/j.autcon.2025.106034

work page arXiv 2024
[51]

A. K. Avgoren, Enhancing IFC Model Interpretability Using Knowledge Graphs and Large Language Models with Integrated Visual Support, Master’s thesis, Technical University of Munich, 2025. URL:https://mediatum.ub.tum.de/ node?id=1782062

2025
[52]

BIM Dictionary, buildingSMART Data Dictionary (bSDD),https:// bimdictionary.com/en/buildingsmart-data-dictionary/1, 2025. 49

2025
[53]

N. Wang, R. Issa, C. J. Anumba, A framework for intelligent building information spoken dialogue system (ibisds), in: EG-ICE 2021 workshop on intelligent computing in engineering, Universitätsverlag der TU Berlin, 2021, p. 228. URL:https://www.researchgate.net/publication/353742039_A_ Framework_for_Intelligent_Building_Information_Spoken_Dialogue_ System_iBISDS

work page arXiv 2021
[54]

N. Wang, R. R. Issa, C. J. Anumba, Nlp-based query-answering system for information extraction from building information models, Journal of comput- ing in civil engineering 36 (2022) 04022004. doi:10.1061/(ASCE)CP.1943-5487. 0001019

work page doi:10.1061/(asce)cp.1943-5487 2022
[55]

N. Wang, R. R. Issa, C. J. Anumba, Transfer learning-based query classification for intelligent building information spoken dialogue, Automation in Construc- tion 141 (2022) 104403. doi:10.1016/j.autcon.2022.104403

work page doi:10.1016/j.autcon.2022.104403 2022
[56]

Krijnen, J

T. Krijnen, J. Beetz, A sparql query engine for binary-formatted ifc building models, Automation in Construction 95 (2018) 46–63. doi:10.1016/j.autcon. 2018.07.014

work page doi:10.1016/j.autcon 2018
[57]

doi:10.3390/app15147647

B.Liu, H.Chen, Bimcoder: Acomprehensivelargelanguagemodelfusionframe- work for natural language-based bim information retrieval, Applied Sciences 15 (2025) 7647. doi:10.3390/app15147647

work page doi:10.3390/app15147647 2025
[58]

J. Wang, Q. Yang, Y. Chen, A large language model–based approach for au- tomatically optimizing bim, in: 2024 43rd Chinese Control Conference (CCC), IEEE, 2024, pp. 8518–8523. doi:10.23919/CCC63176.2024.10661504. 50

work page doi:10.23919/ccc63176.2024.10661504 2024
[59]

Fuchs, A

S. Fuchs, A. Borrmann, Towards semantic enrichment of ifc models through language modeling (2025). doi:10.17868/strath.00093271

work page doi:10.17868/strath.00093271 2025
[60]

S. Du, Y. Gao, W. Tong, Y. Weng, Towards agent: Llm-based framework for robotic construction by leveraging ifc, in: ISARC. Proceedings of the Inter- national Symposium on Automation and Robotics in Construction, volume 42, IAARC Publications, 2025, pp. 358–365. doi:10.22260/ISARC2025/0048

work page doi:10.22260/isarc2025/0048 2025
[61]

P. Guo, H. Xue, J. Ma, J. C. P. Cheng, Advancing bim information retrieval with an llm-based query-domain-specific language and library code function alignmentsystem, AutomationinConstruction178(2025)106374.doi:10.1016/ j.autcon.2025.106374

work page arXiv 2025
[62]

C. Du, S. Esser, S. Nousias, A. Borrmann, Text2bim: Generating building mod- els using a large language model-based multi-agent framework, arXiv preprint arXiv:2408.08054 (2024). doi:10.1061/JCCEE5.CPENG-6386

work page doi:10.1061/jccee5.cpeng-6386 2024
[63]

Y. Gao, Y. Xiong, X. Gao, K. Jia, J. Pan, Y. Bi, Y. Dai, J. Sun, H. Wang, H. Wang, Retrieval-augmented generation for large language models: A survey, arXiv preprint arXiv:2312.10997 2 (2023). doi:10.48550/arXiv.2312.10997

work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2312.10997 2023
[64]

Webber, A programmatic introduction to neo4j, in: Proceedings of the 3rd annual conference on Systems, programming, and applications: software for humanity, 2012, pp

J. Webber, A programmatic introduction to neo4j, in: Proceedings of the 3rd annual conference on Systems, programming, and applications: software for humanity, 2012, pp. 217–218. doi:10.1145/2384716.2384777

work page doi:10.1145/2384716.2384777 2012
[65]

URL:https://langchain-ai.github.io/ langgraph/

LangChain, Langgraph, 2025. URL:https://langchain-ai.github.io/ langgraph/

2025
[66]

gpt-oss-120b & gpt-oss-20b Model Card

S. Agarwal, L. Ahmad, J. Ai, S. Altman, A. Applebaum, E. Arbus, R. K. Arora, 51 Y. Bai, B. Baker, H. Bao, et al., gpt-oss-120b & gpt-oss-20b model card, arXiv preprint arXiv:2508.10925 (2025). doi:10.48550/arXiv.2508.10925

work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2508.10925 2025
[67]

Groq, Groq.com,https://groq.com, 2025

2025
[68]

Groq, Compound,https://console.groq.com/docs/compound/systems/ compound, 2025

2025
[69]

Donkers, H

S. Donkers, H. Ledoux, J. Zhao, J. Stoter, Automatic conversion of ifc datasets to geometrically and semantically correct citygml lod3 buildings, Transactions in GIS 20 (2016) 547–569. doi:10.1111/tgis.12162

work page doi:10.1111/tgis.12162 2016
[70]

Shang, X

W. Shang, X. Huang, A survey of large language models on generative graph analytics: Query, learning, and applications, IEEE Transactions on Knowledge and Data Engineering (2025). doi:10.1109/TKDE.2025.3609877

work page doi:10.1109/tkde.2025.3609877 2025
[71]

Z. Hong, Z. Yuan, Q. Zhang, H. Chen, J. Dong, F. Huang, X. Huang, Next- generation database interfaces: A survey of llm-based text-to-sql, IEEE Trans- actions on Knowledge and Data Engineering (2025). doi:10.1109/TKDE.2025. 3609486

work page doi:10.1109/tkde.2025 2025
[72]

Zhang, J

Q. Zhang, J. Dong, H. Chen, D. Zha, Z. Yu, X. Huang, Knowgpt: Knowledge graph based prompting for large language models, Advances in Neural Infor- mation Processing Systems 37 (2024) 6052–6080. doi:10.48550/arXiv.2312. 06185

work page doi:10.48550/arxiv.2312 2024
[73]

Y. Choi, J. Baek, S. J. Hwang, System prompt optimization with meta-learning, arXiv preprint arXiv:2505.09666 (2025). doi:10.48550/arXiv.2505.09666. 52

work page doi:10.48550/arxiv.2505.09666 2025
[74]

Lamsal, M

R. Lamsal, M. R. Read, S. Karunasekera, Crisistransformers: Pre-trained language models and sentence encoders for crisis-related social media texts, Knowledge-Based Systems 296 (2024) 111916

2024
[75]

Lamsal, M

R. Lamsal, M. R. Read, S. Karunasekera, Semantically enriched cross-lingual sentence embeddings for crisis-related social media texts, arXiv preprint arXiv:2403.16614 (2024)

work page arXiv 2024
[76]

G. N. Lilis, G. I. Giannakis, D. V. Rovas, Detection and semi-automatic cor- rection of geometric inaccuracies in ifc files, in: 14th International Conference of IBPSA-Building Simulation 2015, BS 2015, Conference Proceedings, IBPSA, 2015, pp. 2182–2189. doi:10.26868/25222708.2015.2581. 53 Appendix A. System Prompt Template You are an expert assistant for...

work page doi:10.26868/25222708.2015.2581 2015
[77]

Use SQLite for element hierarchy, properties, and counts
[78]

Use Neo4j for spatial navigation, adjacency, shortest paths, and connectivity among rooms/doors/stairs
[79]

Neo4j has no labels; always match nodes by properties, e.g., (n {type:‘room’})
[80]

Relationships are CONNECTS with properties distance and type
[81]

Treat all names as strings; use case-insensitive matching when necessary

Showing first 80 references.