Techniques for organizing components in a component management service based on geometry and metadata-based classification
Inventors
Page, Kaustubh • Jahjah, Karl-Alexandre
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12530617-B1
Publication Date
2026-01-20
Expiration Date
Abstract
In example embodiments, a component management service identifies related components using a combination of ML model geometry-based classification and text-based classification. The component management service accesses a plurality of components, wherein each component is associated with a geometry mesh and textual metadata. It classifies each component based on geometric similarity by providing the geometry mesh as input to one or more ML models. A geometry classification confidence score is produced for each classification. The component management service also classifies each component based on textual similarity by providing the textual metadata as input to the one or more ML models. A textual classification confidence score is produced for each classification. It calculates an overall confidence score for each classification by combining the geometry classification confidence score and the textual metadata classification confidence score. It displays components together that have an overall confidence score that exceeds a threshold for a same classification.
Core Innovation
The invention relates to generating a user interface for modeling infrastructure that presents related components usable in an infrastructure model proximate to each other. A component management service accesses a plurality of components, where each component models a portion of the infrastructure and is associated with a geometry mesh and textual metadata. The service classifies components using both geometric similarity and textual similarity, and predicts both a classification and corresponding confidence scores for each classification.
For geometric similarity classification, the service cleans the geometry mesh, extracts features from the cleaned geometry mesh to produce a first vector of features, compares the first vector of features to features of components already grouped together, and predicts a classification and a geometry classification confidence score. For textual similarity classification, the service normalizes the textual metadata, extracts features from the normalized textual metadata to produce a second vector of features, and predicts a classification and a textual metadata classification confidence score.
The service calculates an overall confidence score for each classification based on the geometry classification confidence score and the textual metadata classification confidence score. In response to a key-word entered by a user, the service searches the plurality of components to locate a first component responsive to the key-word, displays the first component in the user interface, and displays one or more other components proximate to the first component that have an overall confidence score exceeding a threshold for the same classification.
Claims Coverage
The independent claims are clm-00001, clm-00010, and clm-00012. They combine dual classification based on geometry mesh and textual metadata, overall confidence scoring, and keyword-responsive display of a first component and proximate or grouped related components exceeding a threshold for the same classification.
Infrastructure model component UI with proximate related components
A method for generating a user interface for modeling infrastructure that presents related components usable in an infrastructure model proximate to each other by accessing a plurality of components usable in the infrastructure model, wherein each component is associated with a geometry mesh and textual metadata; receiving a key-word entered by a user; searching the plurality of components usable in the infrastructure model to locate a first component that is responsive to the key-word; displaying the first component in the user interface; and displaying one or more other components proximate to the first component in the user interface that have an overall confidence score that exceeds a threshold for a same classification as the first component.
Dual classification using geometry mesh and textual metadata
Classifying each component based on geometric similarity by cleaning the geometry mesh, extracting features from the cleaned geometry mesh to produce a first vector of features, comparing the first vector of features to features of components already grouped together, and predicting a classification and a geometry classification confidence score; and classifying each component based on textual similarity by normalizing the textual metadata, extracting features from the normalized textual metadata to produce a second vector of features, and based on the second vector of features predicting a classification and a textual metadata classification confidence score.
Overall confidence score for each classification
Calculating an overall confidence score for each classification based on the geometry classification confidence score and the textual metadata classification confidence score for each classification.
Computing device with geometry and text ML models and confidence calculation module
A computing device configured to generate a user interface for modeling infrastructure that presents related components usable in an infrastructure model proximate to each other comprising a first ML model configured to classify each component based on geometry mesh features to predict a classification and a geometry classification confidence score; a second ML model configured to classify each component based on textual metadata to predict a classification and a textual metadata classification confidence score; a confidence calculation module configured to calculate an overall confidence score for each classification based on the geometry classification confidence score and the textual metadata classification confidence score for each classification; and a user interface module configured to receive a key-word entered by a user, cause a search to locate a first component responsive to the key-word, display the first component, and display one or more other components proximate to the first component with an overall confidence score that exceeds a threshold for a same classification as the first component.
Non-transitory device-readable medium for dual classification and keyword UI presentation
A non-transitory electronic device readable medium storing instructions operable to access a plurality of components usable in an infrastructure model, wherein each component is associated with a geometry mesh and textual metadata; classify each component based on geometric similarity to predict a classification and a geometry classification confidence score; classify each component based on textual similarity to predict a classification and a textual metadata classification confidence score; calculate an overall confidence score for each classification based on the geometry classification confidence score and the textual metadata classification confidence score; receive a key-word entered by a user; search the plurality of components to locate a first component responsive to the key-word; display the first component in a user interface; and display one or more other components grouped together with the first component in the user interface that have an overall confidence score that exceeds a threshold for a same classification as the first component.
Across clm-00001, clm-00010, and clm-00012, the claims cover a modeling user interface generation workflow that performs geometry-mesh-based and textual-metadata-based classification with confidence scoring, combines confidence into an overall confidence score per classification, and uses a user key-word to display a responsive component and proximate or grouped related components whose overall confidence exceeds a threshold for the same classification.
Stated Advantages
Enables displaying related components proximate to a first keyword-responsive component that exceed a threshold for the same classification.
Uses both geometry classification confidence and textual metadata classification confidence to calculate an overall confidence score for each classification.
Documented Applications
A user interface for modeling infrastructure presenting related components usable in an infrastructure model proximate to each other, in response to a key-word entered by a user.
Interested in licensing this patent?