1.
Andreas Blumauer
CEO & Managing Partner
Semantic Web Company /
PoolParty Semantic Suite
Semantic AI
Bringing Machine Learning, NLP
and Knowledge Graphs together

2.
Introduction
Semantic Web Company (SWC)
▸ Founded in 2004, based in Vienna
▸ Privately held
▸ 50 FTE
▸ Software Engineers & Consultants for
NLP, Semantics and Machine learning
▸ Developer & Vendor of
PoolParty Semantic Suite
▸ Participating in projects with €2.5
million funding for R&D
▸ ~30% revenue growth/year
▸ SWC named to KMWorld’s
‘100 Companies That Matter in
Knowledge Management’ in 2016,
2017 and 2018
▸ www.semantic-web.com
2 PoolParty Semantic Suite
▸ Most complete Semantic Middleware
on the Global Market
▸ Semantic AI: Fusion of Knowledge
Graphs, NLP, and Machine Learning
▸ W3C standards compliant
▸ First release in 2009
▸ Current version 7.0
▸ On-premises or cloud-based
▸ Over 200 installations world-wide
▸ Named as Sample Vendor in
Gartner’s Hype Cycle for AI 2018
▸ KMWorld listed PoolParty as
Trend-Setting Product 2015, 2016,
2017, and 2018
▸ www.poolparty.biz

3.
Agenda
3
Semantic
AI
▸ Introduction to Semantic AI
▹ Current status of Artificial
Intelligence
▹ Machine Learning, NLP and
Knowledge Graphs coming
together
▹ Six core aspects of Semantic AI
▸ Use Cases

4.
SEMANTIC
ARTIFICIAL INTELLIGENCE
#SemanticAI: Bringing Machine Learning,
NLP and Knowledge Graphs together
4

5.
Predictions and
ground truths
about AI
Herbert A. Simon (1965)
▸ AI pioneer Herbert A. Simon (Carnegie Mellon University) had predicted in 1965 that
"machines will be capable, within twenty years, of doing any work a man can do"
Doug Lenat (1998)
▸ “Those of us who lack shared knowledge and experience often don't understand
much of what they say to each other. In the same way, today's computers, which
don't share even the common-sense knowledge we all draw on in our everyday
speech and writing, can't comprehend most of our speech or texts.”
5
HAL 9000 (1968) KITT (1982) Siri (2011)

6.
AI suffers from
a lack of
common sense,
but ...
6
Google Assistant (2018)

7.
… is talented in
solving isolated
problems based
on isolated data
sets
7
Monte Carlo TS
Deep Learning
Deep Learning
Genetic Algorithms
Neuronal networks
Case based reasoning
Face recognition Game AI Fraud detection

8.
Gartner Hype
Cycle for Artificial
Intelligence, 2018
“The rising role of
content and context for
delivering insights with AI
technologies, as well as
recent knowledge graph
offerings for AI
applications have pulled
knowledge graphs to the
surface.”
8

9.
What makes
someone an
intelligent
being?
Assessment of
the current status
of Artificial
Intelligence
Level Example Questions
(6)
Create
How to convert an inefficient AI system
architecture to a more efficient one
by replacing your choice of components?
How would you improve …?
Can you formulate a theory for …?
Can you predict the outcome if …?
(5)
Evaluate
Which kinds of knowledge models
are best for machine learning, and why?
What is your opinion of …?
How would you prioritize …?
What would you use to support the view …?
(4)
Analyse
How does a graph database
and a semantic knowledge model
work together?
How is ... related to ...?
What is the function of ...?
What conclusions can you draw ...?
(3)
Apply
How can taxonomies be used
to enhance machine learning?
Why is … significant?
How is … an example of …?
What elements would you use to change …?
(2)
Understand
What is the difference between
an ontology and a taxonomy?
What is the difference between …?
What is the main idea of …?
Which statements support …?
(1)
Remember
Who is the inventor of
the World Wide Web?
Who is …?
Where is …?
Why did …?
9
Bloom’s Taxonomy: Classify cognitive processes

10.
Remember
10
Perth
Australia
Perth is one of the
most isolated
major cities in the
world, with a
population of
2,022,044 living
in Greater Perth.
Australia is a
member of the
OECD, United
Nations, G20,
ANZUS, and the
World Trade
Organisation.
Country
City
is a
is a
is located in
Avoid illogical answers:distance between
Commonwealth
of Nations
International
Organisation
is part of
is a
Support complex Q&A:
Which cities located in the
Commonwealth of Nations
have a population of more
than 2 mio. people?
“Knowledge graphs
silently accrue ‘smart
data’ — i.e., data that can
be easily read and
‘understood’ by AI
systems.”
Gartner Hype Cycle for
Artificial Intelligence,
2018

11.
Remember
Knowledge Graphs
& Knowledge
Extraction
Knowledge Graphs (KG) can cover
general knowledge (often also
called cross-domain or
encyclopedic knowledge), or
provide knowledge about special
domains such as biomedicine.
In most cases KGs are based on
Semantic Web standards, and have
been generated by a mixture of
automatic extraction from text or
structured data, and manual
curation work.
Examples:
▸ DBpedia
▸ Google Knowledge Graph
▸ YAGO
▸ OpenCyc
▸ Wikidata
11 Who is the inventor of the World Wide Web?

12.
“Understand”
Google Featured
Snippets based on
Sentence
Compression
Algorithms
To train Google’s artificial Q&A brain, the
company uses old news stories, where
machines start to see how headlines
serve as short summaries of the longer
articles that follow. But for now, the
company still needs its team of PhD
linguists.
Spanning about 100 PhD linguists across
the globe, the Pygmalion team produces
“the gold data,” while the news stories
are the “silver.” The silver data is still
useful, because there’s so much of it. But
the gold data is essential.
WIRED article
12 What is the difference between an ontology and a taxonomy?

13.
“Create”
Example for
DL-based
“creativity” Aiva’s compositions still require human input with regards to
orchestration and musical production. In fact, Aiva’s creators envisage
a future where man and machine will collaborate to fulfill their
creative potential, rather than replace one another.
http://www.aiva.ai/
13 After having listened to a large
amount of music and learned its own
models of music theory, Aiva
composes its very own sheet music.
These partitions are then played by
professional artists on real
instruments in a recording studio,
achieving the best sound quality
possible.

14.
AI Methods -
An overview
14
Artificial
Intelligence (AI)
Artificial Neural
Network (ANN)
Symbolic AI
(GOFAI*)
Sub-Symbolic AI Statistical AI
Knowledge graphs &
reasoning
Natural Language
Processing (NLP)
Machine Learning
* Good old-fashioned AI
Word Embedding
(Word2Vec)
Deep Learning
(DNN)
Natural Language
Understanding
Entity Recognition
& Linking
Knowledge
Extraction
Semantic enhanced
Text Classification

15.
Observations from
the AI Market
Lack of AI Governance
▸ Companies have concerns about validity, explainability and unintended bias of AI.
Lack of AI Strategy
▸ Many organizations are currently undertaking POCs from a large pool of AI vendors only for
tactical benefits.
Low Data Quality & Data Silos
▸ 80% of the work of data scientists is acquiring and preparing data. A demon that can drive up that
80% and often makes initiatives impossible are data silos.
Danger of Vendor Lock-in
▸ Use of black-boxes instead of hybrid middleware approaches connecting internal training assets
to third-party machine-learning solutions.
Lack of Knowledge
▸ Only 1 in 10 enterprises feel they have a competent approach to mining data, which ultimately
hampers AI efforts.
▸ A shortage of AI skills and risk managers' lack of familiarity with the technology increase the risk.
▸ Gartner (March 2018): “Clarify Strategy and Tactics for Artificial Intelligence by Separating Training and Machine Learning” by
Anthony Mullen, Magnus Revang, Erick Brethenoux
▸ Harvard Business Review (2016): “Breaking Down Data Silos” by Edd Wilder-James
▸ Gartner (April 2018): “CIOs Can Manage the Risks of AI Investments” by Jorge Lopez, Paul E. Proctor
15

16.
Six Core Aspects
of Semantic AI
> Read more
16 1. Data Quality
Semantically enriched data
serves as a basis for better
data quality and provides
more options for feature
extraction.
2. Data as a Service
Linked data based on W3C
Standards can serve as an
enterprise-wide data
platform and helps to provide
training data for machine
learning in a more
cost-efficient way.
3. No black-box
Semantic AI ultimately
leads to AI governance
that works on three
layers: technically,
ethically, and on the
legal layer.
4. Hybrid approach
Semantic AI is the
combination of methods
derived from symbolic AI
and statistical AI. It is not
only focused on process
automation, but also on
intelligence augmentation.
5. Structured data
meets text
Most machine learning
algorithms work well either
with text or with structured
data. Semantic AI is based on
entity-centric data models.
6. Towards self
optimizing
machines
ML can help to extend
knowledge graphs, and in
return, knowledge graphs
can help to improve ML
algorithms.

17.
Data Quality
Training data is
semantically
enriched with help
from semantic
knowledge models
17
PoolParty Semantic Classifier combines machine learning algorithms
(SVM, Deep Learning, Naive Bayes, etc.) with Semantic Knowledge Graphs.

18.
Benchmarking
the PoolParty
Semantic
Classifier
Improvement of
5.2% compared
to traditional
(term-based)
SVM
18
Features used Classifier F1 (5-fold) Variance
Terms LinearSVC 0.83175 0.0008
Concepts from REEGLE + Shadow Concepts LinearSVC 0.84451 0.0011
Concepts from REEGLE LinearSVC 0.84647 0.0009
Terms + Concepts from REEGLE + Shadow Concepts LinearSVC 0.87474 0.0009
Reegle thesaurus
A comprehensive SKOS taxonomy
for the clean energy sector
(http://data.reeep.org/thesaurus/guide)
● 3,420 concepts
● 7,280 labels (English version)
● 9,183 relations (broader/narrower + related)
Document Training Set
1,800 documents in 7 classes
Renewable Energy, District Heating Systems,
Cogeneration, Energy Efficiency, Energy (general),
Climate Protection, Rural Electrification

19.
Data as a
Service
Proposal for a
Cognitive
Computing
Platform
Architecture
19 Unstructured Data
Structured Data
Knowledge Graphs
Machine
Learning
Semantic
Layer
Cognitive
Applications

20.
No black-box
Explainable AI
(XAI)
Explaining an image classification prediction made by Google’s Inception network, highlighting
positive pixels. The top 3 classes predicted are “Electric Guitar” (p = 0.32), “Acoustic guitar” (p =
0.24) and “Labrador” (p = 0.21)
From: Ribeiro, M. T., Singh, S., & Guestrin, C. (2016, August). Why should i trust you?: Explaining the predictions of any classifier. In
Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining (pp. 1135-1144). ACM.
20 Explainable AI as an AI whose
decision-making mechanism for
a specific problem can be
understood by humans who have
expertise in making decisions for
that specific problem.
Explainable AI has been used for years in AI
that are based on transparent methods.
These include Expert Systems or Symbolic
Reasoning Systems - anything that is
considered GOFAI (Good Old-Fashioned AI)
methods)

21.
No black-box
Explainable AI
(XAI)
Explaining a text classification prediction made by PoolParty Semantic Suite, highlighting positive
concepts and terms.
21

22.
Hybrid Approach
22
Artificial Intelligence
ANN
Symbolic AISub-Symbolic AI Statistical AI
KR & reasoning
NLP
Machine Learning
Word Embedding Deep Learning
Natural Language
Understanding
Entity Recognition &
Linking
Knowledge Extraction
Semantic enhanced
Text Classification
In Semantic AI, various methods from
Symbolic AI are combined with
machine learning methods, and/or
neuronal networks.
Examples:
● Semantic enrichment of
text corpora to enhance
word embeddings
● Extraction of semantic features
from text to improve ML-based
classification tasks
● Combine ML-based with
Graph-based entity extraction
● Knowledge Graphs as a Data
Model for Machine Learning
● ….

23.
Structured data
meets text
Knowledge Graphs
as a Data Model for
Machine Learning
These transformations can result in loss of information and introduce bias. To solve this problem, we
require machine learning methods to consume knowledge in a data model more suited to represent
this heterogeneous knowledge. We argue that knowledge graphs are that data model.
Three examples for the benefits of using knowledge graphs:
▸ they allow for true end-to-end-learning,
▸ they simplify the integration of heterogeneous data sources and data harmonization,
▸ they provide a natural way to seamlessly integrate different forms of background knowledge.
Wilcke X, Bloem P, De Boer V. The Knowledge Graph as the Default Data Model for Machine Learning. Data Science. 2017 Oct 17;1-19.
Available from, DOI: 10.3233/DS-170007
23 Traditionally, when faced with
heterogeneous knowledge in a
machine learning context, data
scientists preprocess the data
and engineer feature vectors so
they can be used as input for
learning algorithms (e.g., for
classification).

24.
Towards self
optimizing
machines
The Singularity is
just around the
corner ;-)
Mike Bergman (2014): Knowledge-based Artificial Intelligence
24

25.
USE CASES
Bringing Machine Learning, NLP and
Knowledge Graphs together
25

26.
The LinkedIn
Economic Graph
26
The LinkedIn Economic Graph is a
digital representation of the global
economy based on
▸ 560 million members,
▸ 50 thousand skills,
▸ 20 million companies,
▸ 15 million open jobs, and
▸ 60 thousand schools.
https://economicgraph.linkedin.com/
“LinkedIn has a vast quantity of data.
While much of the data is
structured—graph nodes and edges,
normalized fields in database
records—a great deal of it is simply
natural language text. Attaching
structure and meaning to this text is
essential to LinkedIn’s overall
mission of connecting its members
to opportunity.”

27.
27
13485 skills /
competences
2942
occupations
For more information
go to:
https://ec.europa.eu/
esco/portal/home
EventAdvisor
helps to 'putting
out feelers' to find
the right people
and interesting
content

28.
Deep Text
Analytics
Combining
ML-based and
Graph-based text
mining with rules
Bain Capital is a
venture capital
company based in
Boston, MA.
Since inception it has
invested in hundreds
of companies. In 2018,
Bain had $75b AUM.
Graph-based
Entity
Extraction
ML-based
Entity Extraction
Regular
Expression-based
Extraction
SHACL-based
Rules Engine
Give me all paragraphs in
documents talking about American
Private Equity firms with AUM
higher than $20b
28

29.
▸ Data Quality & Date Governance
▹ Content aboutness in a defined
framework
▹ Data relationships and context within a
unified organizational model
▹ Connections across disparate datasets
▸ Improved Machine Learning
▹ Hierarchical or other mapped
relationships allow for recommending
similar content when exact matches not
found
▹ Granularity allows for more specific
recommendations
▹ Consistency across structure results more
precise analysis and predictions
Source: Suzanne Carroll, Data Science Product Director at XO Group
Why Data
Scientists need
Semantic
Models
29

30.
Examples for
enterprise use
cases based on
(Semantic) AI
30

31.
One question
at the end
Will Artificial
Intelligence
make
Subject Matter
Experts
obsolete?
31 Imagine you want to
build an application
that helps to identify
patients and
treatments pairings.
Which will you prefer?
1. Solely based on machine learning,
2. based on doctors' knowledge only,
3. or a combination of both?

32.
Thank you for
your interest!
Andreas Blumauer
CEO, Semantic Web Company
▸ Mail andreas.blumauer@semantic-web.com
▸ Company https://www.semantic-web.com
▸ LinkedIn https://www.linkedin.com/in/andreasblumauer
▸ Twitter https://twitter.com/semwebcompany
32
© Semantic Web Company - http://www.semantic-web.com and http://www.poolparty.biz/