Voronoi Treemap Force-Directed Layout for Interactive Data Labeling

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing interactive data labeling systems fail to provide a dynamic and adaptable method for clustering data items, requiring significant time and effort to prepare and label groupings, and lack the ability to allow data items to be located anywhere on the display without additional parameters, with fixed or limited dynamic grouping and positioning options.

Innovation Solution

The implementation of a force-directed power diagram technique in Voronoi treemaps enables users to interactively arrange and group objects by dragging and dropping cells, allowing for predictable and stable layout adjustments, enabling free positioning and dynamic grouping without additional parameters, and supporting side-by-side comparison and focus+context interactions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If typical visual interactive labeling pipeline methods are used, then users can arrange and group objects to construct mental models, but the process requires significant time and effort to individually identify and label large amounts of data

Engineering Contradiction:
Improveinteractivity of arranging and groupingVSAvoidtime to label data
Core Design Contradiction:
Ease of operationVSLoss of time

Solution Approach 1:

The system automatically computes similarity measures between objects and performs clustering to generate groupings, allowing the system to serve itself rather than requiring manual arrangement of all objects. Users only need to interact when they want to refine or adjust the automatically generated groupings.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system allows users to adjust similarity thresholds and clustering parameters to control the level of automation. By changing these parameters, users can balance between automatic grouping (faster) and manual control (more precise), adapting the system to different stages of the labeling process.

Inventive Principle:
Principle #35Parameter changes

2Adaptability or versatility

If explicit grouping by enclosure or similarity is used, then grouping intent is recorded and space efficiency is improved, but additional parameters are required to determine categories and encode visual distinctions

Engineering Contradiction:
Improvegrouping flexibilityVSAvoidparameters required for grouping
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system extracts the complexity of determining grouping parameters by automatically computing similarity measures and determining optimal cluster configurations. This removes the burden from users of having to manually specify categories, encoding schemes, and visual distinctions, while still providing explicit groupings with recorded intent.

Inventive Principle:
Principle #2Taking out (Extraction)

3Ease of operation

If data items are allowed to be located anywhere on the display, then free positioning and direct manipulation are enabled, but layout stability and predictability become difficult to maintain

Engineering Contradiction:
Improvedirect manipulation capabilityVSAvoidlayout stability
Core Design Contradiction:
Ease of operationVSStability of the object's composition

Solution Approach 1:

The system implements dynamic layout adjustment where the positions of objects and groups are continuously optimized based on similarity relationships. When objects are moved or groups are modified, the layout automatically reconfigures to maintain stability while preserving user-initiated changes, balancing freedom of manipulation with overall layout coherence.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS11222453B1Interactive visual data categorization systems and methods
Publication Date: 2022.01.11 HAMAD BIN KHALIFA UNIVERSITY
  • US11222453B1 patent drawing
  • US11222453B1 patent drawing
  • US11222453B1 patent drawing

AI summary

The present disclosure provides interactive visual data labeling systems and methods that improve upon power-diagram Voronoi Treemaps by supporting interactive arranging and grouping of objects represented by thumbnail images. The provided interactive visual data labeling system enables new and innovative fluid interactions and direct manipulation of treemap cells to arrange objects and groups of objects, and create, merge, expand, or collapse groups. In effect, these enabled fluid interactions and direct manipulation provide an intuitive, stable, and predictable Voronoi treemap for users. In at least some aspects, the provided visual data labeling system uses a force-directed power diagram technique, which enables controlling the stability of the Voronoi treemap layout and updating it smoothly when dragging a cell across a parent cell. In effect, the predictability and stability enables the Voronoi treemap to react as a user expects during the user's interaction with the treemap.