Building Information Model Feedback Validation for Reliable Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing information models for buildings require reliable and accurate training, but user feedback can be incorrect, leading to worse model predictions and the need for improved feedback assessment methods.
Innovation Solution
A method and system for assessing user feedback on building information models by comparing it with historical feedback to identify contradictions and filter out incorrect input, using a large language model (LLM) like ChatGPT to generate and refine the model iteratively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If user feedback is directly used to train the information model, then the model can be refined and improved, but incorrect feedback will lead to worse model predictions
Solution Approach 1:
The patent introduces an intermediary assessment module that sits between the user feedback input and the model training process. This intermediary component evaluates and validates user feedback before it is used for training, filtering out incorrect feedback while allowing valid feedback to proceed to model refinement. The intermediary acts as a gatekeeper that prevents harmful incorrect feedback from directly affecting model predictions.
Solution Approach 2:
The patent implements a feedback mechanism where user feedback is assessed and used to refine the information model iteratively. The system receives feedback, assesses its validity, and incorporates valid feedback to improve model predictions. This feedback loop enables continuous improvement of model accuracy while the assessment component ensures only quality feedback influences the model.
2Productivity
If feedback assessment is performed manually, then the process is simple to implement, but it requires significant time and labor resources
Solution Approach 1:
The patent replaces manual feedback assessment with an automated assessment module that uses computational methods to evaluate user feedback. Instead of human reviewers manually checking feedback, the system uses automated algorithms to assess feedback validity, significantly reducing the time and labor required while maintaining consistent assessment quality.
Solution Approach 2:
The feedback assessment system performs self-service by automatically evaluating and validating feedback without requiring external manual intervention. The system autonomously processes feedback, determines its validity, and prepares it for model training, eliminating the need for manual assessment operations and reducing overall processing time.
3Ease of operation
If all user feedback is accepted without assessment, then the system is easy to operate, but the model training reliability deteriorates
Solution Approach 1:
The patent introduces an intermediary assessment layer between feedback input and model training. Users can provide feedback in a simple manner without needing to understand the assessment process, while the intermediary component automatically evaluates and filters feedback before it reaches the training system. This maintains ease of operation for users while ensuring training reliability through automated validation.
Solution Approach 2:
The system implements a feedback mechanism where user input is automatically assessed and validated before being used for training. The feedback loop includes an assessment stage that ensures only reliable feedback influences model training, maintaining both user-friendly operation and training accuracy through automated validation processes.
Data Source
AI summary
A computer implemented method for assessing a feedback of an information model of a building, comprising: receiving raw data of a building; generating an information model of a building based on the raw data; receiving a feedback of a user for the generated information model; assessing the received feedback and determining an assessing result; providing the assessing result for further processing.