A Brief Introduction to Knowledge Graphs

A Brief Introduction to
Knowledge Graphs
Taxonomy Boot Camp London
16 October 2019
Presented by
Heather Hedden
Hedden Information Management

▪ Taxonomy consultant
– Independent, through Hedden Information Management
– Previously as an employed and contract consultant
▪ Formerly staff taxonomist
– At various companies: Gale/Cengage Learning, Viziant, First Wind
▪ Instructor of online and onsite taxonomy courses
– Independently through Hedden Information Management
– Previously at Simmons University - Library & Information Science School
▪ Author of The Accidental Taxonomist (2010, 2016, Information Today, Inc.)
About Heather Hedden
2
© 2019 Hedden Information Management

Overview
▪ Knowledge graphs are gaining more interest.
▪ There is some uncertainty in how to definite knowledge graphs.
▪ Knowledge graphs span the fields of knowledge management information
science, information technology, computer science.
▪ There are increasing applications of knowledge graphs.
▪ Knowledge graphs closely relate to ontologies and often include taxonomies.
Introduction to Knowledge Graphs
3© 2019 Hedden Information Management

What knowledge graphs are
▪ The organization and representation of a knowledge base as a graph:
as a network of nodes and links, not a table of rows and columns
▪ Usually based on data in graph databases, rather than relational databases
▪ Is both human-readable and machine-readable
▪ Usually includes, but not limited to, visualizations, such as of…
– A display of interconnected nodes and links
– A display of related information in a "fact box“
– An output of graph analytics

Knowledge Graphs
7
Knowledge graph example
https://commons.wikimedia.org/wiki
/File:Wikidata-knowledge-graph-
madame-x-2019.png
Fuzheado [CC BY-SA 4.0
(https://creativecommons.org/licens
es/by-sa/4.0)]

Issues with knowledge graph definitions
▪ “Knowledge graphs” have different meanings from different perspectives:
from knowledge engineers, data engineers, data architects, etc.
▪ Sometimes considered the same as a knowledge base, or at least a
knowledge base that is represented as a graph.
▪ Wikipedia redirects “Knowledge graph” to “Ontology (information science).”
▪ An entire presentation and article on definition issues:
"Towards a Definition of Knowledge Graphs," by Lisa Eherlinger and Wolfram
Wöß, CEUR Workshop Proceedings presentation at SEMANTiCS 2016
http://ceur-ws.org/Vol-1695/paper4.pdf
Defining Knowledge Graphs

Issues with knowledge graph definitions
Related to ontologies, whose definition is also not precise. An ontology can be:
1. A complex knowledge organization system with defined types (classes or
individuals), attribute properties and semantic relations
2. A semantic layer, in accordance with W3C standards, that defines the generic
types, attributes and relations, and can be applied to taxonomies and other
knowledge organization systems
▪ For definition #2 of an ontology, the combination of the generic semantic-
layer ontology along with the specific instances (such as found in a
taxonomy), then is something else, called a knowledge graph.
▪ But the combination may also called an ontology, resembling definition #1,
which results in the conflation of “ontology” and “knowledge graph.”

Knowledge graphs and ontologies
▪ “A knowledge graph acquires and integrates information into an ontology and
applies a reasoner to derive new knowledge.” - Eherlinger and Wöß, “Towards
a Definition of Knowledge Graphs.”
▪ Whereas an ontology can be a generic model template of how things are
related to each other, a knowledge graph is the actual instance of that model.
▪ A knowledge graph is an ontology + instance data (instance terms and links to
data and content)
➢ Knowledge graphs are ontologies and more.
➢ A knowledge graph may also comprise multiple ontologies, or an ontology and
other vocabularies.

Creating knowledge graphs
▪ Create or utilize taxonomies, apply ontologies, and link to data/content.
▪ Follow SKOS, OWL, and RDF standards of the W3C.
– For example, all nodes must have URIs (Uniform Resource Identifies).
▪ Graph-database software tools can help.
▪ Data may be added manually or automated/minded, or a combination.
Manual technique is similar to that for creating taxonomies and ontologies,
including:
▪ Inventory of content and data
▪ Development of use cases
▪ Mapping relationships

Knowledge graphs and ontologies both:
▪ Represent nodes (things) and relationships between them
▪ Can be visually represented in the same way of nodes and defined
relationships and then may look the same in the visualization
▪ Are based on Semantic Web standards, such as RDF triples
▪ Tend to have been more the expertise domain of computer scientists and data
scientists of than information professionals/taxonomists, but that’s changing!
Also a growing area of interest in knowledge management (business).
Knowledge Graphs and Ontologies

Knowledge graphs and ontologies are based on RDF
RDF, a standard model for data interchange on the Web, uses URIs to name
things and the relationship between things, which are referred to as triples:
(1) Subject – (2) Predicate – (3) Object.
Subject Predicate Object
SUBJECT PREDICATE
OBJECT
Knowledge Graphs and Ontologies
14
Rome, Italy ItalyCapCityof
CapCity
Rome, ItalyItaly

Knowledge graphs and knowledge organization systems
(taxonomies, thesauri, ontologies, etc.)
▪ Knowledge graphs may comprise multiple domains and thus multiple
knowledge organization systems.
▪ Knowledge graphs can link together disparate sources of vocabularies and
data.
Knowledge Graphs and Knowledge Organization Systems

What knowledge graphs can do
▪ Integrate knowledge
▪ Serve data governance
▪ Provide semantic enrichment
▪ Bring structured and unstructured data together
▪ Provide unified view of different kinds of unconnected data sources
▪ Provide a semantic layers on top of the metadata layer
▪ Improve search results beyond machine learning and algorithms
▪ Answer complex user questions instead of merely returning documents on a
topic
▪ Combine with deep text analytics, semantic AI, and machine learning
Uses, Implementations, and Examples

Implementations of knowledge graphs
▪ Recommendation engine (such as in ecommerce)
▪ Expert finder
▪ Question-answering based on data
▪ Enterprise knowledge management
▪ Search and discovery
▪ Customer 360 – view of everything known about customers
▪ Compliance
Implementation usually requires:
▪ a content management system
▪ search engine

Examples of implementations
▪ Search engine results
– Google’s Knowledge Graph (since 2012)
Freebase, a proprietary graph database acquired by Google in 2010 when
it bought Metaweb
– Microsoft’s Satori (since 2012)
Microsoft Research’s Trinity graph database and computing platform
▪ Healthdirect Australia - public website health symptom checker
https://www.healthdirect.gov.au/symptom-checker
combining data on initial symptoms, gender and age, and then questions on
proposed additional symptoms

Taxonomies in Support of Search
19
Knowledge Graphs from
Google searches

Implementations of knowledge graphs
Companies that have built knowledge graphs
Airbnb
Alibaba
Amazon
Apple
Bank of America
Bloomberg
Facebook
Now smaller, medium-sized companies are also building knowledge
graphs.
20
Genentech
Goldman Sachs
JPMorgan Chase
LinkedIn
Microsoft
Uber
Wells Fargo

Questions/Contact
21
Heather Hedden
Taxonomy Consultant
Hedden Information Management
Carlisle, MA USA
+1 978-467-5195
www.hedden-information.com
accidental-taxonomist.blogspot.com
www.linkedin.com/in/hedden
Twitter: @hhedden

A Brief Introduction to Knowledge Graphs

More Related Content

What's hot

Similar to A Brief Introduction to Knowledge Graphs

More from Heather Hedden

Recently uploaded

A Brief Introduction to Knowledge Graphs