Back to Wiki

[107] Using Text Understanding to Create Formatted Semantic Web from BIM

Seed Paper BIM Classification & Semantic Enrichment 2023

Paper Description extracted from PDF · abstract · 150 words

The application of BIM in the building life cycle needs to be continuous. The information collected and accumulated in the early stages should flow to the subsequent phases. However, BIM applications currently focus on collision inspection, compliance inspection, and engineering calculation, few models can be successively used in the following stages. Remodeling is required in the operation and maintenance period, resulting in waste. Meanwhile, some of the information accumulated by BIM might be frequently used in the operation and maintenance stage, while some data are relatively rarely used. The semantic web can help manage building information at all stages. But the generation of a semantic web is mostly manually completed. It is necessary to standardize the repeated semantic description in the model and convert BIM into a standard semantic model for information indexing, reducing the resource consumption of model loading and optimizing the efficiency of the operation and maintenance system.

AuthorsJingming Li
Year2023
VenueComputational Design and Robotic Fabrication
DOI10.1007/978-981-19-8637-6_17
Source DatabaseSeed (CumInCAD)
Bridge-to-GNNCategory Seed
GNN ArchitectureKnowledge Graph Embedding (inferred from title)
Graph EncodingBIM Element Topology Graph (inferred from title)
AEC TaskSemantic Enrichment (inferred from title)
CohortConsolidation (2023)

Extracted Methodology heuristic — verify before citing

Implementation FrameworkPyTorch, Grasshopper / Rhino, Revit / Dynamo, OWL / SPARQL (heuristic — verify)
Primary MetricAccuracy (heuristic — verify)
Primary Value60% (heuristic — verify)
Key Findingmethod of using Natural Language Processing (NLP) to understand the text and infer the relationship between entities according to the knowledge map (heuristic — verify)
Quality Assessment1/6 · rigor: Low ·
Baseline: ✓ (heuristic — verify)

Detected Technical Stack auto-detected from PDF · regex catalogue match

GNN architectures detectedKnowledge Graph Embedding (TransE/RotatE/DistMult)
Frameworks / librariesPyTorch · Grasshopper / Rhino · Revit / Dynamo · RDF / OWL / SPARQL · OpenAI / GPT API · Sentence Transformers
Reported metrics + valuesAccuracy: 60%
Quality (heuristic, 6-flag)0/6 · rigor: Low
Code/Data: —Benchmark: —Baseline: —CV/Split: —Ablation: —Reproducible: —

Lineage Families

llm-kg-hybrid (auto-suggested)

Enabled By / Precursors

None tracked

Extended By

None in corpus

Infrastructure Dependencies

BIM (auto-suggested) Knowledge Graph (RDF/OWL/SPARQL) (auto-suggested) PyTorch (detected from PDF) Grasshopper / Rhino (detected from PDF) Revit / Dynamo (detected from PDF) RDF / OWL / SPARQL (detected from PDF) OpenAI / GPT API (detected from PDF) Sentence Transformers (detected from PDF)

Research Frontier Tags

bim-graph

Part of the GML/GNN in AEC Systematic Review (PRISMA 2020) — 112 papers, 2020–2026