4. Find the right thing
Knowledge graphs in search
Get the best
summary
Go deeper and
broader
Source: https://blog.google/products/search/introducing-knowledge-graph-things-not/
11. Idea originating from the semantic web, now more linked
to enterprise data integration and machine learning
Graph-based data model, useful when the domain evolves,
or new data sources are added to the graph
11
13. Orders of magnitude higher scales than
GOFAI knowledge bases
Simple knowledge representation, no
formal semantics
Vocabulary reuse, networks of small
modular vocabularies
Incomplete, inconsistent, always
changing
Built via human-AI pipelines (w/ ETL,
information extraction etc.)
Many large open-source projects with
strong communities
Knowledge graph services for developers
Graph: 238m publications, 151m authors
Graph: 97m things Developer resources
Developer resources
15. Content-wise some overlap, but
different paradigm: knowledge
engineering, decentralized data
publishing i.e., identifiers, reusable
schemas, interlinking
More related to semantic networks,
frames, rule-based NLP than LLM
Knowledge graphs are a common
source of embeddings in AI systems
18. The knowledge graph
Statements, items, properties
Item id’s start with a Q, property id’s start with a P
18
Q84
London
Q334155
Sadiq Khan
P6
head of government
19. The knowledge graph
Items can be classes, entities, values
19
Q7259
Ada Lovelace
Q84
London
Q334155
Sadiq Khan
P6
head of government
Q727
Amsterdam
Q515
city
Q6581097
male
Q59360
Labour party
Q145
United Kingdom
20. The knowledge graph
Statements have context
Statements may include context
• Qualifiers (optional)
• References (required)
Two types of references
• Internal, linking to another item
• External, linking to webpage
20
Q84
London
Q334155
Sadiq Khan
P6
head
of government
9 May 2016
london.gov.uk
21. The knowledge graph
Co-edited by bots and humans
Human editors can register or work anonymously
Community creates bots to automate routine tasks
23k active human users, 340+ bots
26. Making assumptions
explicit, documenting use
cases “ontology cards”
Predicting costs,
understanding the most
resource-intensive parts of
the process
Aligning ontology
engineering tasks to
motivations of participants,
incentives design
37. Reasoning and
planning
• Huge collections of machine-readable knowledge
• Testbed for argumentation approaches
Natural language
processing,
computer vision
• Knowledge graph embeddings
Visualization
• Challenges due to scale, incompleteness, noise, frequent
changes
• Visualization as a tool to understand knowledge
engineering communities
Accessibility • Graphs are used to organised multimedia content, can
we make them more accessible?
TAS Hub • Rich source of declarative knowledge, facilitates
explanations
38. Thank you
Elena Simperl
@esimperl
Publications on Scholar and DBLP
Thanks to: David Abian, Kholoud Al
Ghamdi, Gabriel Amaral, Paul Groth,
Jonathan Hare, Lucie Kaffee, Laura Koesten,
Elisavet Koutsiana, Eddy Maddalena, Albert
Merono, Chris Phetean, Alessandro
Piscopo, Odinaldo Rodrigues, Miaojing Shi,
Pavlos Vougiouklis