# GeoGraphViz: Geographically Constrained 3D Force-Directed Graph for Knowledge Graph Visualization

**Authors:** Sizhe Wang, Wenwen Li, Zhining Gu  
**arXiv:** [2304.09864](https://arxiv.org/abs/2304.09864)  
**Date:** 2023-04-14  
**Summary page:** https://www.summarizepaper.com/en/paper/2304.09864v1/

## Key points

- Knowledge graphs are important for linking and integrating cross-domain data, concepts, tools, and knowledge to enable data-driven analytics.
- Google's knowledge graph contains 5 billion entities and over 500 billion facts, supporting over a third of the 100 billion queries made to Google each month.
- The National Science Foundation (NSF) has started a new program called Convergence Accelerator to develop cutting-edge knowledge graph technologies for building an open knowledge network that fosters convergence research.
- KnowWhereGraph is a knowledge graph that connects environmental datasets related to natural disasters, agriculture, and soil properties to understand the environment's impacts on society.
- Challenges arise in accessing and understanding massive datasets for non-developers as well as knowledge engineers.
- Existing popular graph visualization tools mainly support 2D visualization without geographical reference for geospatial data while traditional GISs provide powerful solutions but significantly alter local clusters' patterns.
- Wang et al developed a novel 3D map-based graph visualization algorithm that enables interactive exploration of graph content and patterns spatially.

## Summary

The use of knowledge graphs has become increasingly important in linking and integrating cross-domain data, concepts, tools, and knowledge to enable data-driven analytics. Google's knowledge graph alone contains 5 billion entities and over 500 billion facts, supporting over a third of the 100 billion queries made to Google each month. In the geospatial domain, progress has been made in creating data and metadata schemas to annotate data in the broad Earth and environmental science domains. The National Science Foundation (NSF) has also started a new program called Convergence Accelerator to develop cutting-edge knowledge graph technologies for building an open knowledge network that fosters convergence research. One key advancement in spatial sciences is the creation of KnowWhereGraph, a knowledge graph that connects environmental datasets related to natural disasters, agriculture, and soil properties to understand the environment's impacts on society. However, challenges arise for non-developers as well as knowledge engineers to access and understand massive graph-ready datasets. As human brains are especially good at visual inspection, providing an intuitive presentation of the knowledge graphs is critical for understanding the graph's entities and interconnections among them. Existing popular graph visualization tools mainly support 2D graph visualization with nodes representing entities and edges representing relationships between them. However, these solutions lack geographical reference for geospatial data such as hurricane tracks or food supply chain interruptions caused by wildfires near farmlands. Although location information may be encoded in the graph as text, few tools support visualizing this special information in a virtual geographical space like a map. Traditional GISs provide powerful solutions for overlaying multiple geospatial data on a basemap using flow maps to support graph data visualization. However, these map-based visualizations alter local clusters' patterns significantly as strengths of edges connecting nodes are not considered. To tackle these challenges effectively, Wang et al developed a novel 3D map-based graph visualization algorithm that enables interactive exploration of graph content and patterns spatially. The algorithm extends a 3D force-directed graph by integrating a web map, an additional geolocational force, and a force balancing variable that allows for dynamic adjustment of the 3D graph structure and layout. This mechanism helps create a balanced graph view between the semantic forces among the graph nodes and the attractive force from geolocation to a graph node. Their solution offers a new perspective in visualizing and understanding spatial entities and events in a knowledge graph. In conclusion, knowledge graphs are crucial for linking cross-domain data to enable data-driven analytics. However, challenges arise in accessing and understanding massive datasets. Graph visualization tools mainly support 2D visualization without geographical reference for geospatial data while traditional GISs provide powerful solutions but significantly alter local clusters' patterns.