Two-Mode Network Visualization for Subnetwork Pattern Detection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing data visualization systems are inadequate for analyzing two-mode networks, as they fail to effectively identify and visualize subnetwork patterns, especially in large networks with weighted links, due to limitations in scalability and the inability to detect high-level patterns involving overlapping nodes.

Innovation Solution

A method and system for visualizing search results in two-mode networks by detecting subnetwork patterns and generating interactive visualizations that separate and connect different types of entities, providing information about relationships and allowing for user interaction to refine the visualization.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If related art visualization techniques for single mode networks are used, then general network exploration is enabled, but identification of subnetwork patterns in two-mode networks becomes difficult

Engineering Contradiction:

Improvenetwork exploration capabilityVSAvoidsubnetwork pattern identification

Core Design Contradiction:

Adaptability or versatilityVSDifficulty of detecting and measuring

Solution Approach 1:

The visualization system segments the two-mode network into distinct regions representing different node types (e.g., authors and documents), with dedicated space for displaying subnetwork patterns. This segmentation allows the system to maintain adaptability for general network exploration while specifically enhancing subnetwork pattern identification through structured visual organization.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system adds a visual dimension by creating a specialized layout that displays subnetwork patterns as separate visual entities within the network graph. This dimensional enhancement transforms the abstract concept of subnetwork patterns into visually detectable structures, resolving the contradiction between general exploration capability and specific pattern identification difficulty.

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

2Difficulty of detecting and measuring

If biclustering visualization techniques are used for two-mode networks, then subnetwork patterns can be illustrated, but scalability and generality for weighted networks are reduced

Engineering Contradiction:

Improvesubnetwork pattern visualizationVSAvoidscalability and generality

Core Design Contradiction:

Difficulty of detecting and measuringVSAdaptability or versatility

Solution Approach 1:

The visualization system implements a universal framework that can handle both weighted and unweighted two-mode networks, as well as networks of varying sizes. The system uses general-purpose graph visualization techniques combined with pattern detection algorithms that work across different network types, thereby achieving both effective subnetwork pattern visualization and broad scalability.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system dynamically adjusts visualization parameters such as node size, edge thickness, and pattern highlighting intensity based on network characteristics including weight values and network size. This parameter adaptability allows the same visualization framework to effectively display subnetwork patterns across diverse network types while maintaining scalability.

Inventive Principle:

Principle #35Parameter changes

3Productivity

If related art pattern finding algorithms are used, then basic pattern detection is achieved, but manual inspection is required and high-level patterns are missed

Engineering Contradiction:

Improvepattern detection capabilityVSAvoidhigh-level pattern information

Core Design Contradiction:

ProductivityVSLoss of information

Solution Approach 1:

The visualization system implements interactive feedback mechanisms where users can explore detected patterns, drill down into details, and adjust detection parameters. This feedback loop allows the system to iteratively refine pattern detection results and reveal high-level patterns that span multiple basic patterns, eliminating the need for manual inspection while preserving comprehensive pattern information.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system merges multiple basic pattern detection results to identify higher-level patterns that emerge from the combination of overlapping subnetworks. By integrating detection results across different patterns and visualizing their intersections, the system recovers high-level pattern information that would be lost in isolated basic pattern detection.

Inventive Principle:

Principle #5Merging (Combining)

Data Source

PatentUS11048713B2System and method for visual exploration of search results in two-mode networks

Publication Date: 2021.06.29 FUJIFILM BUSINESS INNOVATION CORP

AI summary

A method of visualizing search results is provided. The method includes receiving a content feature, detecting, within a network, a subnetwork pattern representing a relationship satisfying a condition and associated with an entity of a first or a second type, the entity being associated with the content feature, and generating a visualization based on the detected subnetwork pattern. The visualization includes a first region representative of the first type of entity, a second region representative of the second type of entity, and a linking region connecting the first region to the second region and providing information about the represented relationship.