KnowWhereGraph: A Cross-Domain Geo-Knowledge Graph and Service Stack

*KnowWhereGraph: A Cross-Domain Geo-Knowledge Graph and Service Stack*

May 6, 2022We announce/Vienna/, the first public release of the 
KnowWhereGraph Project (KWG;https://knowwheregraph.org/ 
<https://knowwheregraph.org/>). KWG is an open cross-domain knowledge 
graph and associated toolset that rapidly raises the situational 
awareness of data scientists and decision makers by providing detailed 
area briefings for any place on Earth within seconds to answer questions 
such as:

  *      "What is here?"
  *      "What happened here before?"
  *      "Who knows more?"
  *      "How does it compare to other regions or previous events?"
  *

KWG features services for representing, exploring, visualizing, and 
analyzing data at the interface between humans and their environment 
that are open, cross-domain, deeply integrated, and densely connected. 
At the project's core is the KnowWhere*Graph*, a geo-knowledge graph 
that is based on existing standards like RDF, OWL, SHACL, SSN/SOSA, 
OWL-Time, and GeoSPARQL, and also incorporates custom ontologies. 
KnowWhereGraph uses the S2 hierarchical discrete global grid (DGG) for 
spatial representations and inference 
(https://stko-kwg.geog.ucsb.edu/lod/ontology 
<https://stko-kwg.geog.ucsb.edu/lod/ontology>).

In a nutshell, KWG is a gazetteer of gazetteers. It contains more than 
ten different types of region identifiers (e.g., ZIP codes, FIPS codes, 
administrative boundaries, climate zones, named places) and links those 
to over 20 data layers about transportation, health, climate, disasters, 
soil and agriculture, and so on. Because many phenomena, such as storms, 
floods, or earthquakes, do not respect human-made boundaries, users can 
represent any region of their interest with a collection of S2 cells 
(currently at level 13). While KWG aims to serve global data, most of 
the current layers are US-centric. Please 
seehttps://www.knowwheregraph.org/graph/ 
<https://www.knowwheregraph.org/graph/>for an overview table. KWG also 
links out to other graphs.
Its current size exceeds 12.5 billion triples (graph statements) with 
industry and non-governmental organization pilots in disaster relief, 
agricultural land use, applications in credit and risk evaluations, and 
food-related supply chains. KWG data include observations of past and 
present natural hazards (e.g., hurricanes, wildfires, landslides, smoke 
plumes) and spatial characteristics related to climate (e.g., 
temperature, precipitation, air quality), soil properties, 
crop/land-cover types, demographics (e.g., health data about obesity and 
diabetes), experts (and their expertise), and transportation 
infrastructure, among many others. Essentially, we are centrally 
incurring the high cost of data integration to substantially reduce the 
time spent on data wrangling for all end users.

Altogether, KWG and its supporting toolset enable queries such: "For a 
given earthquake simulation, show me the population at risk based on 
some demographic criteria, nearby major transportation/evacuation 
infrastructure, highlight soils at risk of liquefaction, and propose 
experts familiar with the region, e.g., in terms of local health matters."

The graph can be queried directly using the SPARQL querying interface 
athttps://stko-kwg.geog.ucsb.edu/sparql 
<https://stko-kwg.geog.ucsb.edu/sparql>, or browsed with the*Knowledge 
Explorer*athttps://stko-kwg.geog.ucsb.edu 
<https://stko-kwg.geog.ucsb.edu/>. The Knowledge Explorer helps users 
focus their data search by providing a set of filters corresponding to 
data characteristics, and hyperlinked results allow the user to navigate 
and de-reference the graph. Together, these tools help reveal the 
graph's content and structure.
We are leveraging the graph with additional tools being developed for 
specific contexts. The open-source*Geo-Enrichment Toolbox plugins*hosted 
on both ArcGIS and QuantumGIS software enable Geographic Information 
Systems (GIS) users to easily incorporate a growing variety of natural 
hazards, climate, and socioeconomic data from KWG into their own spatial 
analysis. To assist humanitarian response efforts following a natural 
disaster, the*GeoGraphVis*tool provides situational awareness, 
visualizing, for example, physical properties of a hurricane in relation 
to health characteristics of populations in its path.

In combination with the Knowledge Explorer, the*Expert Similarity 
Search*helps responders to identify people with expertise relevant to 
the disaster situation quickly.

The*Crop Impact Assessment*tool enhances strategic planning during 
disasters by providing online analysis, forecasting, and alerts to 
ensure key stakeholders throughout the supply chain are ready with 
backup strategies to keep products moving. It also allows farmers and 
growers to identify mitigation strategies and build resilience in the 
face of such events.

The National Science Foundation funds KWG as part of its Convergence 
Accelerator program's*Open Knowledge Network (OKN)*initiative. The team 
includes members from academia (University of California, Santa Barbara; 
Kansas State University; Michigan State University; Arizona State 
University; University of Southern California), the nonprofit sector 
(Direct Relief); industry (Esri; Oliver Wyman; Hydronos Labs), and the 
US federal government (US Geological Survey; US Department of Agriculture).

At this early stage in the public release of KWG, we hope to collect 
constructive feedback and anticipate downtimes due to substantial query 
load./Vienna/is a rolling release, so we will fix issues (and add new 
one ;-)) as we receive feedback. Developing and deploying a 12B 
triples-large graph by integrating 30 complex data layers is one thing; 
keeping such a graph stable and running is a very different story. We 
would love to hear and learn from you about your experience in scaling 
and sustaining such graphs and about your ideas for new data and denser 
links among datasets.*KWG and OKN are community efforts; please get 
involved.*After browsing the graph and trying the tools, please consider 
answering a few questions in the survey 
athttps://forms.gle/Z8WU17vizatHdD2g8 
<https://forms.gle/Z8WU17vizatHdD2g8>. Thanks for your time and interest.

/Finally, this release is dedicated to our team member E. Lynn Usery 
(USGS), who passed away in March 2022, and our advisory board member 
Peter A. Fox (RPI), who passed away in March 2021. Both were very 
prominent, long-term supporters of the geo-semantics, spatial data 
science, and knowledge graph communities and inspired all of us. They 
left a deep mark on the broader community. Their contributions and 
kindness shall not be forgotten./