Semantic Classification of BIM Entities Using Geometric Probability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current CAD technologies face challenges in efficiently and accurately classifying entities in building information models (BIMs) without semantic annotation, requiring labor-intensive domain-specific rule sets and manual intervention.
Innovation Solution
A computer-implemented method and system for semantic classification of BIM entities using a two-step process: determining initial probability distributions based on geometric information and updating them with relative geometric information, allowing for automated, context-dependent classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If domain-specific rule sets are used for semantic classification, then classification accuracy can be improved, but the complexity of the system increases and requires labor-intensive design and maintenance
Solution Approach 1:
The patent replaces manual domain-specific rule sets with a machine learning-based automated classification system. The system uses trained models that automatically learn classification patterns from data, eliminating the need for manual rule creation and maintenance while maintaining high classification accuracy.
Solution Approach 2:
The classification system performs self-service by automatically updating and refining its own classification rules through machine learning. The system continuously improves its accuracy by learning from new data without requiring external intervention to create or maintain rule sets.
2Productivity
If automated classification algorithms are implemented, then productivity increases, but classification reliability may deteriorate due to incorrect classifications
Solution Approach 1:
The patent implements feedback mechanisms where classification results are continuously evaluated and used to refine the machine learning models. This feedback loop ensures that automated classifications maintain high reliability by correcting errors and improving accuracy over time through systematic learning from outcomes.
3Adaptability or versatility
If comprehensive rule sets are created to handle diverse object types, then adaptability improves, but the time and resources required for rule set maintenance increase
Solution Approach 1:
The patent replaces manual rule set creation and maintenance with automated machine learning systems that can adapt to diverse object types. The system learns classification patterns automatically from training data, providing comprehensive coverage across different building elements without requiring manual intervention for each new object type.
Data Source
AI summary
The current invention concerns a computer-implemented method, a computer system, and a computer program product for the semantic classification of an entity in a building information model (BIM). The BIM comprises multiple target entities. Update data is obtained. For each target entity, geometric information about the target entity is obtained from the BIM. For each target entity, an initial probability distribution of semantic classification is determined based on the obtained geometric information about the target entity. Relative geometric information about the target entities is obtained from the BIM. For each target entity, an updated probability distribution of semantic classification is determined based on the obtained relative geometric information, the initial probability distributions of all target entities, and the update data. For each target entity, a semantic classification is selected based on the updated probability distribution of the target entity.