Introduction to Knowledge Graphs

When data learn to think outside the box

Andreas Blumauer CEO of Company Andreas Semantic Web serves customers Blumauer Company >200 graduates

founded Founder & CEO 2004 of parent company of Vienna active at Director of headquartered located

PoolParty Software developer & version 7.0 Ltd London vendor of

Gartner Introduction awarded based on by KMWorld

used for manages integrates Text Mining CMS/DMS/DA with M/.. part of based on

Search engine Ontologies standard for part of

Knowledge Graphs standard for Taxonomies

© Semantic Web Company 2019 2 Why Knowledge Graphs?

WHAT’S THE PROBLEM?

Many Challenges in (Enterprise) Information Management & AI

© Semantic Web Company 2019 3 “Search” is still about documents only … and in enterprises it’s painful

© Semantic Web Company 2019 4 Different shades of metadata

© Semantic Web Company 2019 5 User-agnostic and context-free data models Do machines understand user intent? Do they have enough context?

French SUV Which Sport-utility vehicle from France provides enough space for my 1. Intent recognition family with 3 kids? 2.

3. Background knowledge

Car Loadspace Max no. of seats The Peugeot The Kia 5008 breaks Sportage is new ground as a KIA Sorento 550 litres 7 our sporty large SUV with SUV packed many features. Peugeot 5008 823 litres 7 with smart features BMW X3 550 litres 5

© Semantic Web Company 2019 6 Commonwealth is a International of Nations Organisation

is part of Support complex Q&A:

Australia is a Country Which cities located in the Commonwealth of Nations is located in have a population of more Artificial than 2 mio. people? “Intelligence” Perth is a City Avoid illogical answers: Background distance between knowledge is key Perth is one of the Australia is a most isolated member of the major cities in the OECD, United world, with a Nations, G20, population of ANZUS, and the 2,022,044 living World Trade in Greater Perth. Organisation.

© Semantic Web Company 2019 7 Machine Learning per Data Silo

Deep Learning Monte Carlo TS Neuronal Lack of AI Strategy Genetic Algorithms Deep Learning networks Case based reasoning

© Semantic Web Company 2019 8 Who is CEO of a Bank, headquartered in Europe that generates revenue per employee higher than 400,000 Euro?

Desktop Data Integration

Knowledge workers spend Head- Ticker Revenue a large part of their quarter Symbol The task awaiting Tidjane working time on ad hoc 48,200 Thiam when he takes over Credit Zurich VTX: CSGN CHF 23.4b from Brady Dougan as the data integration tasks, the Suisse so-called "research". 266,273 new chief executive at Credit Suisse Group AG is HSBC London LON: HSBA USD 60.0b clear: how to pull the Swiss Which parts of it could be 147,425 bank out of a Allianz Munich ETR: ALV EUR 122.3b automated? post-financial crisis rut. 101,104 Deutsche Frankfurt ETR: DBK EUR 33.5b Bank

© Semantic Web Company 2019 9 Head- Ticker Umsatz quarter Symbol The task awaiting Tidjane Thiam when he takes over Credit Zürich VTX: CSGN CHF 23.4b from Brady Dougan as the Suisse new chief executive at Many challenges in HSBC London LON: HSBA USD 60.0b Credit Suisse Group AG is clear: how to pull the Swiss bank out of a Allianz München ETR: ALV EUR 122.3b data management post-financial crisis rut.

▸ Heterogeneous data Deutsche Frankfurt ETR: DBK EUR 33.5b Bank ▸ Implicit ▸ No background knowledge 48,200 ▸ Data quality difficult to measure ▸ Proprietary schemas 266,273 ▸ Unstructured data ▸ Ambiguity 147,425 ▸ Multilinguality ▸ Various units and currencies 101,104 → From Search to QA engines

© Semantic Web Company 2019 10 The Free Universal Construction Kit is a matrix of adapter bricks that enable complete interoperability between ten popular children's construction toys.

Free Universal Construction Kit (Golan Levin et al) © Semantic Web Company 2019 11 ‘Interoperability’ begins with people Bringing various types of mindsets together

Data-centric people Knowledge-centric people

Service-oriented industries Asset-oriented industries

Databases & Excel Content & documents

Structured data Unstructured data

Linked Data Knowledge Graph

Algorithms Collaborative knowledge management

Predictability and automation Product innovation and risk mitigation

Ontologies and Machine Learning Taxonomies and Collaboration

“Actionable Data!” “Better Decisions!”

Understand the customer and market place better Stay or become the expert in the field

12 What is a Knowledge Graph?

From a Knowledge From a Data Architect’s From a Data Engineer’s Engineer’s perspective perspective perspective A Knowledge Graph is a model of Structured as an additional It provides a structure and a knowledge domain created by virtual data layer, the KG lies on common interface for all of your subject-matter experts with the top of existing databases or data data and enables the creation of help of intelligent machine sets to link all your data smart multilateral relations learning algorithms. together at scale – be it throughout your databases. structured or unstructured.

© Semantic Web Company 2019 13 Other Graphs than Knowledge Graphs Mind the gap!

Social Graphs / Networks Wordnets

14 ▸ Neptune ▸ Marklogic ▸ AllegroGraph ▸ GraphDB ▸ Azure Cosmos DB ▸ Stardog Oracle Spatial & Graph Neo4j Which Graph ▸ ...

Model? Property Graph RDF Graph (Triple Stores)

RDF versus Main use case Traverse a graph Query a graph Property Graphs Typical applications Path Analytics, Data Integration, → Converging approaches Social Network Analysis Knowledge Representation

Standards No standards W3C Semantic Web standards → Gremlin, Cypher, → SPARQL 1.1 PGQL, ...

Additional options Shortest path Inferencing (Example) calculations

© Semantic Web Company 2019 15 Sounds interesting? Gimme some examples!

WHO USES KGs? An overview over typical KG use cases

© Semantic Web Company 2019 16 Can you see the Knowledge Graph?

Taobao (Alibaba) PoolParty GraphViews

© Semantic Web Company 2019 17

© Semantic Web Company 2019 18 Knowledge Graph Adoption

Source: Collapsing the IT Stack: Clearing a path for AI adoption Alan Morrison (Sr. Research Fellow at PwC) 19 Knowledge Graphs for Information Retrieval Semantic tagging, query expansion, faceted search, classification, similarity-based recommender

Now I can find all the documents in one place, and get assistance with it. Metadata harmonization Semantic tagging Entity linking Semantic Search Index

20 Example healthdirect Australia

https://www.healthdirect.gov.au/

© Semantic Web Company 2019 21 Semantic Search

PowerTagging and PowerSearch for SharePoint

© Semantic Web Company 2019 22 As a researcher in pharmaceutical industry, I want to plan new experiments more efficiently. → Linking Structured to I want to know what’s already available. Unstructured Data and I’m interested in former experiments Use Case where to Industry Knowledge Graphs Research in ● certain genes were tested ● under specific treatment conditions Life Sciences ● in a target therapeutic area ● with help from categorisation systems like ‘disease hierarchies’

UniProt, ChEMBL

Experiments Documentation

MeSH

DrugBank

© Semantic Web Company 2019 23 Experiments Reusing Graphs

Knowledge Graphs used as a Semantic Search Index

Gartner - Hype Cycle for AI (2018): “Knowledge Graphs serve as means to enrich unstructured information to provide a rich set of additional access points to document repositories”

© Semantic Web Company 2019 24 Knowledge Graphs for Data Integration & Analytics Metadata enrichment, , text mining, entity-centric search, agile reporting

Employee database Enterprise Knowledge Resumes Graph

Labour market statistics

RDF Graph Database

Now I can identify employees along many PoolParty GraphSearch dimensions.

25 Entity-centric Search & Analytics

HR Analytics Dashboard

© Semantic Web Company 2019 26 Knowledge Graphs for E-commerce/Content Mgmt. Personalization, recommender / Dynamic menus, campaigns / Translating jargon into client’s needs

My latest Campaigns Personalization

Rainy Summer Warm Winter Muddy Spring

Rules engine

Mud Techlite Water proof Multi-Grip Multi-Heat Multi-Tech Rain

Super Boots Winter Leather Comfortable Neoprene warm

27 Product Recommender

Wine and Cheese Harmonizer

© Semantic Web Company 2019 28 Example Electronic Arts

Semantics at Play: Electronic Arts' Linked Data Journey

© Semantic Web Company 2019 29 Knowledge Graphs for Virtual Assistants Integrating Semantics into Dialog Workflows

My uncle lives in the same household. Darf ich seine Betreuungskosten absetzen?

teyze ortak Ange- hörige ev

haushalts- zugehöriger uncle Angehöriger

same household

30 Knowledge Graphs for KM Systems Personalization, recommender, matchmaking, push services, smart assistants

31 Use Case Semantic Matchmaking

Event Advisor at SEMANTiCS

© Semantic Web Company 2019 32 I am interested in ...

Architecture Languedoc-Roussillon

Use Case Mediterranean Languedoc is a significant producer Sea of wine, and a major contributor to the surplus known as the "wine Recommender Wine Tasting lake". Today it produces more than a third of the grapes in France, and System is a focus for outside investors.

The region contains the historic cities of Carcassonne, Toulouse, Connecting Montpellier, countless Roman ▸ content to content, monuments, medieval abbeys, Romanesque churches, and old ▸ people to content, Occitanie, vineyard, sea, church, village castles. ▸ people to people.

PowerTagging Thesaurus Server

© Semantic Web Company 2019 33 Example

EIP on Water John Smith John Smith

At the center of the marketplace is the matchmaking function John Smith John Smith

John Peter Smith

https://www.eip-water.eu/

© Semantic Web Company 2019 34 Keynote

How semantic technologies can support us in making better business decisions

https://2019.semantics.cc/keynote-iii-0

© Semantic Web Company 2019 35 Deep Text Analytics ML-based Annotation, Extraction, Classification, Rules Entity Recognition

▸ Corpus statistics / Word embeddings Bain Capital is a → Keyphrase extraction venture capital ▸ Graph-based annotation company based in → Entity/Concept linking Boston, MA. Since inception it has ▸ Corpus Statistics embedded in graphs Graph-based → Shadow Concepts Entity invested in hundreds of Linking companies. In 2018, ▸ Machine-learning-based annotation Bain had $75b AUM. Regular → Named entity recognition (NER) Expressions-based ▸ Machine-learning based classification Annotation → Document Classification

▸ Annotation based on rules Give me all paragraphs in → Regular expressions Semantic documents about “US based Private Equity Rules Engine firms with AUM higher than $20B”

36 TheCompany 12 Use Case Contract Intelligence

Semantic analysis and pre-selection of relevant clauses in contracts

Termination by TheCompany for

© Semantic Web Company 2019 37 Semantic AI

Towards an Explainable AI. Fusing Knowledge Graphs, Machine Learning and NLP

Knowledge Graphs: The Third Era of Computing (Dan McCreary)

© Semantic Web Company 2019 38 Artificial Intelligence (AI)

Symbolic AI Sub-Symbolic AI Statistical AI Fusing (GOFAI*)

Symbolic & Artificial Neural Knowledge graphs & Machine Learning Statistical AI Network (ANN) reasoning

Word Embedding Deep Learning Knowledge (Word2Vec) (DNN) Extraction

Natural Language Entity Recognition Semantic enhanced Understanding & Linking Text Classification

Natural Language Processing (NLP) * Good old-fashioned AI

© Semantic Web Company 2019 39 Semantic Classifier

ML based on semantically enriched training data

PoolParty Semantic Classifier combines machine learning algorithms (SVM, Deep Learning, Naive Bayes, etc.) with Semantic Knowledge Graphs.

© Semantic Web Company 2019 40 Use Case Synthetic data generation

Use KGs to generate test data for your ML (which are GDPR compliant)

© Semantic Web Company 2019 41 KG creation based on Machine Learning The Knowledge Engineer’s perspective

Taxonomy & Entity Extractor & Data Integration & Ontology Server Semantic Classifier Data Linking Controlled vocabularies as a basis for highly precise and text classification

Unstructured Semi-structured Structured Data Data Data Bain Capital is a venture capital company based in Boston, MA. Factsheet Since inception it has invested in hundreds of companies including Schema mapping based AMC Entertainment, Brookstone, on ontologies and Burger King. The company was co-founded by Mitt Romney. Entity linking based on Knowledge Graph UnifiedViews

Identify new candidate concepts to be included in a controlled RDF PoolParty vocabulary Graph Database GraphSearch

42 Which banks are headquartered in Zurich?

Demo QA based on KGs

QAnswer is a so called "knowledge (or ontology) based" QA system

© Semantic Web Company 2019 43 Types of solutions based on Knowledge Graphs

Data Content Knowledge

Process Automation Customer Experience Knowledge Management ▸ Enhanced Machine Learning ▸ Recommender Systems ▸ Graph Analytics ▸ Text Mining & NLP ▸ SEO ▸ Personalization/Skills Mgmt. ▸ Data Classification ▸ Smart Helpdesk Solutions ▸ Knowledge Discovery Portals

Front-end ▸ Document & Text Validation ▸ Virtual assistants/ Q&A engines ▸ Semantic Search

Agile Data Integration Information Management ▸ Linked Data ▸ Semantic Content ▸ Taxonomy Management ▸ Integrating Management ▸ Ontology Management heterogeneous data ▸ Metadata Management ▸ Knowledge Graph Mgmt. Back-end ▸ Entity Linking ▸ Masterdata Management ▸ Data Visualization

44 Anatomy of a Knowledge Graph

THINGS, NOT STRINGS

From Taxonomies over Ontologies to Knowledge Graphs

© Semantic Web Company 2019 45 Things and URIs This is not yet a Knowledge Graph!

St. Mark’s Square http://my.com/2 http://my.com/1 Venice

▸ Learn from text http://my.com/3 corpora ▸ Co-occurences Peggy and word embeddings Guggenheim ▸ Extract entitities via ML Museum ▸ Import CSV/Excel

46 Labels and basic relations This is the backbone of a Knowledge Graph

Venice St. Mark’s Square prefLabel prefLabel Piazza altLabel San Marco related Piazza broader prefLabel

related Town Square altLabel

prefLabel ▸ Import known taxonomies Peggy ▸ Harvest from Guggenheim Knowledge Graphs like Museum DBpedia ▸ Ingest from own DBs/XML

47 Classes and specific relations Ontologies give your knowledge graph an additional "dimensionality"

Venice St. Mark’s Square prefLabel prefLabel containedInPlace Piazza altLabel San Marco Piazza image broader prefLabel

Town Square containedInPlace Monday through altLabel Sunday, all day opening Hours

prefLabel http://schema.org/City ▸ Reuse existing ontologies http://schema.org/TouristAttraction Peggy ▸ Interlink them Guggenheim with your http://schema.org/ArtGallery taxonomies Museum ▸ Map them to http://schema.org/containedInPlace your DBs/XML

48 Metadata and Graph annotations Enrich and qualify data in your knowledge graph

Venice St. Mark’s Square prefLabel prefLabel containedInPlace Piazza CC BY-SA 3.0 altLabel San Marco Piazza image broader prefLabel

Town Square containedInPlace Monday through altLabel Sunday, all day opening Hours Now open ▸ Reuse existing http://schema.org/City schemas ▸ Make use of http://schema.org/TouristAttraction Peggy SPARQL and Guggenheim reasoning http://schema.org/ArtGallery ▸ Make use of Museum constraint http://schema.org/containedInPlace languages like SHACL 49 Graph linking Additional facts for your knowledge graph, nearly for free!

Venice St. Mark’s Square prefLabel prefLabel containedInPlace Piazza CC BY-SA 3.0 altLabel San Marco Piazza image broader prefLabel

Town Square containedInPlace Monday through altLabel Sunday, all day opening Hours Now open

http://schema.org/City ▸ Make use of Named Graphs http://schema.org/TouristAttraction Peggy ▸ Use Machine Guggenheim Learning for http://schema.org/ArtGallery automatic Museum linking ▸ Track data http://schema.org/containedInPlace provenance

50 Link all your data! Bring structured/unstructured enterprise data into the knowledge graph

Venice St. Mark’s Square prefLabel prefLabel containedInPlace Piazza CC BY-SA 3.0 altLabel San Marco Piazza image broader prefLabel R2RML

Town Square containedInPlace Monday through altLabel Sunday, all day opening Hours

Now open The Peggy ▸ Use Text Mining http://schema.org/City Guggenheim ▸ Based on Collection is automatic entity http://schema.org/TouristAttraction Peggy a modern art museum on the extraction Guggenheim Grand Canal in the ▸ Use R2RML http://schema.org/ArtGallery Dorsoduro sestiere of Venice, Italy. Schema to Museum Ontology http://schema.org/containedInPlace mapping ▸ Make use of ML 51 Putting the user into the graph Provide all your data personalized and contextualized

Venice St. Mark’s Square prefLabel prefLabel containedInPlace Piazza CC BY-SA 3.0 altLabel San Marco Piazza image broader R2RML likes prefLabel Town Square containedInPlace Monday through altLabel Sunday, all day visits opening Hours

Now open The Peggy ▸ Build http://schema.org/City Guggenheim recommender Collection is systems http://schema.org/TouristAttraction Peggy a modern art museum on the ▸ Make use of Guggenheim Grand Canal in the graph analytics http://schema.org/ArtGallery Dorsoduro sestiere of Venice, Italy. ▸ Recommender Museum based on http://schema.org/containedInPlace similarity and/or rules 52 Who is CEO of a Bank, headquartered in Europe that generates revenue per employee higher than 400,000 Euro?

Answers instead of search results

Head- Ticker Revenue quarter Symbol The task awaiting Tidjane 48,200 Thiam when he takes over Credit Zurich VTX: CSGN CHF 23.4b from Brady Dougan as the Suisse 266,273 new chief executive at Credit Suisse Group AG is HSBC London LON: HSBA USD 60.0b 147,425 clear: how to pull the Swiss bank out of a Allianz Munich ETR: ALV EUR 122.3b post-financial crisis rut. 101,104 Deutsche Frankfurt ETR: DBK EUR 33.5b Bank

© Semantic Web Company 2019 53 KGs: Linking, Mapping, Reasoning

Who is CEO of a Bank, headquartered in Europe that generates revenue per employee Financial higher than 400,000 Euro? Europe institution part of is a is a

Insurance Switzerland Bank CEO company part of is a is a

HQ in works for Zurich Credit Suisse Tidjane Thiam

Organisation HQ Umsatz The task awaiting Tidjane Thiam when he takes over CS Zürich CHF 23.4b from Brady Dougan as the 48,200 new chief executive at Credit HSBC London USD 60.0b 266,273 147,425 Suisse Group AG is clear: how 101,104 to pull the Swiss bank out of a Allianz München EUR 122.3b post-financial crisis rut.

Deutsche Bank Frankfurt EUR 33.5b

54 Knowledge Graphs support NLU and QA

Who is CEO of a Bank, headquartered in Europe that generates revenue per employee Financial higher than 400,000 Euro? Europe institution part of is a is a

Insurance Switzerland Bank CEO company part of is a is a

HQ in works for Zurich Credit Suisse Tidjane Thiam

Organisation HQ Umsatz The task awaiting Tidjane Thiam when he takes over CS Zürich CHF 23.4b from Brady Dougan as the 48,200 new chief executive at Credit HSBC London USD 60.0b 266,273 147,425 Suisse Group AG is clear: how 101,104 to pull the Swiss bank out of a Allianz München EUR 122.3b post-financial crisis rut.

Deutsche Bank Frankfurt EUR 33.5b

55 Introducing Semantic AI

CHANGE MANAGEMENT

How to implement disruptive technologies in organizations?

© Semantic Web Company 2019 56 C-Level: Knowledge Graphs in various Hype Cycles

Hype Cycle for the Digital Workplace

Hype Cycle for Emerging Technologies Hype Cycle for

© Semantic Web Company 2019 57 Linked Data Life Cycle Data Engineers + Subject Matter Experts + Machine Learning working together

Manual curation Knowledge modelling

Automated Knowledge Graph Development & Machine Learning

58 Computational Linguists Natural languages Schemas/Ontologies

Our approach Ontologists Semantic AI

Data Scientists Statistical models

Taxonomies Taxonomists

© Semantic Web Company 2019 59 Bringing two types of mindsets together

Data-centric people Knowledge-centric people

Service-oriented industries Asset-oriented industries

Databases & Excel Content & documents

Structured data Unstructured data

Linked Data Knowledge Graph

Algorithms Collaborative knowledge management

Predictability and automation Product innovation and risk mitigation

Ontologies and Machine Learning Taxonomies and Collaboration

“Actionable Data!” “Better Decisions!”

Understand the customer and market place better Stay or become the expert in the field

© Semantic Web Company 2019 60 Find the right starting points

© Semantic Web Company 2019 61 Knowledge Graph Management Data Governance along the Linked Data Life Cycle

© Semantic Web Company 2019 62 Semantic Web Starter Kit

Learn more about it!

© Semantic Web Company 2019 63 Make things work

SEMANTIC AI

Build your own applications

© Semantic Web Company 2019 64 Fact sheet: PoolParty Semantic Suite

Most complete and secure Named as Current version Semantic Middleware on 7.0 Representative the Global Market On-premise or Vendor in Gartner’s cloud-based Market Guide for Semantic AI: Hosted AI Services Fusing Graphs, NLP, 2018 and Machine Learning installations Over 200 world-wide KMWorld listed ISO 27001:2013 certified PoolParty as W3C standards compliant Named as Sample Trend-Setting Vendor in Product 2015, 2016, Gartner’s Hype 2017, and 2018 First release in 2009 Cycle for AI 2018 “PoolParty brings the power of Google into the enterprise environment through Knowledge Graphs.”

Knowledge Architect at a Top 3 Oil & Gas company

66 How does it work? The Data Engineer’s perspective

▸ Entity linking PoolParty Unstructured GraphSearch Data ▸ Schema mapping

▸ Data transformation

▸ Data validation

Semi-structured Data ▸ Data cleansing Semantic AI UnifiedViews Knowledge Graph Application ▸ Metadata & Semantic enrichment

Structured Data Database

67 Create unified views of your data!

© Semantic Web Company 2019 68 As an HR manager, for upcoming training programmes, I want to → Linking Structured to identify employees who Unstructured Data ● have a certain skill set ● have a specific degree Example: ● have skills that are increasingly important on the labour market HR Analytics ● fall into a specific salary range

Demo Employee database

Resumes

Labour market statistics

© Semantic Web Company 2019 69 PoolParty Employee Thesaurus Server PoolParty UnifiedViews database

Resumes

Labour market statistics How it works

RDF Graph Database

Now I can identify employees along many

dimensions. PoolParty User PoolParty GraphSearch

© Semantic Web Company 2019 70 Components and Features

© Semantic Web Company 2019 71 PoolParty Components

PoolParty Taxonomy & Ontology Server PoolParty GraphEditor PoolParty UnifiedViews

PoolParty Extractor PoolParty Semantic Classifier PoolParty GraphSearch

© Semantic Web Company 2019 72 SMEs benefiting from Corpus Learning The Knowledge Engineer’s perspective

Taxonomy & Entity Extractor & Data Integration & Ontology Server Semantic Classifier Data Linking Controlled vocabularies as a basis for highly precise knowledge extraction and text classification

Unstructured Semi-structured Structured Data Data Data Bain Capital is a venture capital company based in Boston, MA. Factsheet Since inception it has invested in hundreds of companies including Schema mapping based AMC Entertainment, Brookstone, on ontologies and Burger King. The company was co-founded by Mitt Romney. Entity linking based on Knowledge Graph UnifiedViews

Identify new candidate concepts to be included in a controlled RDF PoolParty vocabulary Graph Database GraphSearch

73 Working with Knowledge Graphs

TECHNICAL DEEP DIVE

Make your data actionable!

© Semantic Web Company 2019 74 Why use Graph Databases?

Dealing with hierarchical and highly connected datasets

Entity-centric views (in contrast to document-centric views)

Exploring the connection between entities of a graph

Integrating heterogeneous data sources

(structured & unstructured in a “schema-late” approach)

© Semantic Web Company 2019 75 What makes a store RDF ready?

It implements at least one of the RDF ready libraries/frameworks: ▸ Eclipse RDF4J ▸ Apache Jena

It offers a SPARQL endpoint to query the database using SPARQL

© Semantic Web Company 2019 76 SPARQL

SPARQL retrieves and manipulates data stored in an RDF graph database

Watch Tutorial from PoolParty Academy

© Semantic Web Company 2019 77 SHACL

Shapes Constraint Language

Data Quality and Consistency: Validating graph-based data against a set of conditions

© Semantic Web Company 2019 78 Concept labels and hierarchic structure

Concept Scheme

skos:hasTopConcept / Example 1 skos:narrower skos:prefLabel label1@en Validating Concept labels and hierarchic structure in a SKOS Concept taxonomy skos:altLabel label2@en Constraints: A concept has to connect to a concept scheme. The labels have to be disjoint and the language must be unique.

© Semantic Web Company 2019 79 Maximum active board members

hasBoardMembership hasBoardMemberStatus Example 2 Legal Entity Board Member Active

Compliance checks

Constraint: There must not be more than two active board members for each Legal Entity.

© Semantic Web Company 2019 80 Consistent geographic information

isLocatedInCountry Country

Example 3 Legal Entity skos:narrower

Data consistency isLocatedInCity City

Constraint: If a Legal Entity has a country and a city assigned, then both must be related with a skos:narrower path, so that the geo information is consistent.

© Semantic Web Company 2019 81 Summarization

TAKE AWAYS

Make your data actionable!

© Semantic Web Company 2019 82 Four-layered Information Architecture

Semantic AI application

Semantic Layer / Knowledge Graph

Metadata Layer

Content & Data Layer

83 From Documents To Entitities

Watch out for ambiguity!

© Semantic Web Company 2019 84 Word embeddings/cooccurences → is related Bla bla bla bla. Bla stove Bla bla bla bla. bla bla Bla bla bla hot ML based on The stove is on. The stove is on. Semantic The stove is hot! The stove is hot! AI Engines Taxonomical model → is-a abstractions Ontological model → reasoning

Separating signal Switched on devices are Switched on dangerous, only if the operating temperature from noise devices are is above 100 degrees and the automatic dangerous shutdown mechanism is devices. broken.

The stove is on. Bla bla bla bla The stove is hot! Bla bla bla bla.

© Semantic Web Company 2019 85 Traditional vs. Graph-based Metadata Management

Traditional approach Show me all documents Graph-based approach about Carnivores

Show me all documents about Carnivore Felidae Keystone Species Keystone Jaguar Species? Cat = Felidae Cat Felidae Carnivore Elephant Keystone Keystone Wolf Species Species Carnivore Rabbit Jaguar Elephant Wolf Rabbit

doc doc doc doc doc doc doc doc

© Semantic Web Company 2019 86 Traditional vs. Graph-based Metadata Management

Traditional approach Show me all documents Graph-based approach about Carnivores

Show me all Metadatadocuments per about KnowledgeCarnivore about Felidae Keystone document metadata Species Keystone Jaguar1. No or little network effectsSpecies? Cat = Felidae 1. Explicit knowledge models Cat Felidae2. No reuse of metadata 2. Reusable and measurable Carnivore3. MetadataElephant resides in silos 3. Metadata is machine-processable Keystone4. DataKeystone quality hard toWolf measure 4. Standards-based metadata Species Species Carnivore Rabbit Jaguar Elephant Wolf Rabbit 5. Not machine-readable 5. Linkable metadata opens silos

doc doc doc doc doc doc doc doc

© Semantic Web Company 2019 87 Next steps

PoolParty Academy Video Tutorials Online Help & Manuals Demo Account

88 Connect

Andreas Blumauer CEO & Managing Partner, Semantic Web Company

[email protected] ▸ https://www.linkedin.com/in/andreasblumauer ▸ https://twitter.com/semwebcompany ▸ https://ablvienna.wordpress.com/

© Semantic Web Company 2019 89