Data-Centric Business Transformation Using Knowledge Graphs

Data-centric architecture in
business transformation
Data Architecture Summit 2018
Chicago, IL
https://www.pwc.com/us/en/services/consulting/technology/emerging-technology.html
October 11, 2018

PwC | Data-Centric architecture in business transformation
Covered in today’s talk
• Introduction: Scoping the opportunity
• Transformation styles and their limitations
• Innovation types
• Deep transformation and a path to AI
• Transformative knowledge graphs
• Conclusion: Pitching to the boss
2PwC | Data-Centric architecture in business transformation 2

Introduction – Scoping the opportunity

Source of a widening corporate inequality gap
Top 20 public/private internet companies by market valuation ($B)
4

Widening corporate inequality examples – Top versus bottom of
global 100
5
337 329
417
559
416 469
725
604
754
851
40 61 69 64 70 81 85 76 88 97
0
200
400
600
800
1000
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
Market caps of top and bottom companies
Number 1 Number 100
8,402
17,438
20,035
0
3,000
6,000
9,000
12,000
15,000
18,000
21,000
24,000
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
Total market cap of Top 100 companies as at 31 March
Apple
Apple
Apple
Apple
AppleApple
AppleExxon
MobilPetroChina
Exxon
Mobil
MarketCap($bn)MarketCap($bn)

Largest changes in market cap by global company cross
industry, 2018
6
(1) Change in market cap from IPO date
(2)Market cap at IPO date
Source: Bloomberg and PwC analysis
Company name Location Industry
Change in market
cap 2009-2018 ($bn)
Market cap 2018
($bn)
1 Apple United States Technology 757 851
2 Amazon.Com United States Consumer Services 670 701
3 Alphabet United States Technology 609 719
4 Microsoft Corp United States Technology 540 703
5 Tencent Holdings China Technology 483 496
6 Facebook United States Technology 383(1) 464
7 Berkshire Hathaway United States Financial 358 492
8 Alibaba China Consumer Services 302(1) 470
9 JPMorgan Chase United States Financials 275 375
10 Bank of America United States Financials 263 307
Known
knowledge
graph builders
Operator of
Taobao and
AliBot
KG builder
v
Known
KG builders

Transformation styles and their limitations

Digital transformation – Data can get buried in the mix
8
Leadership
transformation
Omni-experience
transformation
Worksource
transformation
Operating model
transformation
Information
transformation
• IT strategy and
governance
• Leading in 3D
• Strategic
architecture
• Services
transformation
• Innovation
strategies
• Customer
experience
• Device/mobility
strategies
• Devices: PCs,
mobility,
wearables,
AR/VR
• Social business
• eCommerce
• Vendor sourcing
and management
• IT talent and skills
• Outsourcing
services
• IT organizational
development
• Technology
training
• Enterprise
infrastructure
• AppDev and app
provisioning
• DevOps
• Cloud strategies
• Transformative
tech: IoT, Robotics
and 3D printing
• Enterprise/
nextgen security
• Enterprise
applications
• Information and
data
transformation
• Big data and
analytics
• Cognitive
computing
Research & Advisory Services, IDC, 2018
Digital transformation—IDC’s essential advisory themes

Banks, for example, typically choose one of three digital
transformation directions
9
“Bank to the future: Finding the right path to digital transformation,” PwC FSI White Paper, 2018

Wrap and digitize allows component-by-component
transformation
10
“Bank to the future: Finding the right path to digital transformation,” PwC FSI White Paper, 2018

Bigger picture transformation – Moving to where the new business
will be, with a new data foundation
11
• The shape of digital business is radically different
than what’s come before.
• In order to compete, companies will have to
move to where the new opportunities are.
• Relationship-rich data at scale makes it possible
to get to these opportunities.
• Knowledge graph models as a base for digital
business makes scaling relationship-rich
data possible.

Innovation types

Most innovations are incremental, adding to the stack, with data as
an afterthought (Type I innovation)
13
Hardware
DBMS
OS
Custom code
Hardware
Lots of OSes
1,000+ SQL/
NoSQL DBs
Custom code
ERP+ suites
Hardware
A few more
OSes
More
DBMSes
Custom code
ERP+ suites
Hardware
Lots more OSes
5,000+
databases
Component -
ized suites
Custom code
Cloud layer
Hardware
More types of
OSes
10,000+ DBs +
blockchains
Multicloud
layer
Suites as
services
Various SaaSes
Custom code
Hardware
A few
DBMSes
A few OSes
ERP+ suites
Custom code
Threat of more
application
centric
sprawl
Early1990s Late 1990s 2000s 2010s1973-1990sPre 1970 2020s

Most of the IT workforce just adds to or keeps track of the sprawl
14
0.01
0.9
2
1.6
3.2
US IT workforce in 2016 (in mIllions)
Semantic data
Data related (less semantics)
General
Network/hardware
Software
Total IT workforce = 7.7 million
(= 5 percent of the US overall workforce in 2016)
Sources: US Bureau of Labor Statistics and PwC estimates, 2018

The US as a whole has more opioid abusers than it does IT workers
15
0 50 100 150 200 250 300 350
IT workers
All opioid abusers
Disabled and over 65 years old
Entire US workforce
Entire US population
Number of IT workers in the US in 2016, in context (in millions)
US Census Bureau, Bureau of Labor Statistics, and Health and Department of Human Services, 2018

Object virtualization (Type II) manages complexity, just so IT can
get its arms around the sprawl
16
EnterpriseWeb and PwC, 2015

Type III – Data-centric architecture reduces both application and
database sprawl
17
Applications for execution only Models exposed with the dataTrapped app code and databases
Application centric Data centricversus
Data lake or hub
Semantic model/rules
  Applets

Identify and declare the few hundred business rules you need as
a model
18
“In every company I’ve ever studied, there are only a few hundred key concepts and relationships that the
entire business runs on. Once you understand that, you realize all of these millions of distinctions are just slight
variations of those few hundred important things.”
--Dave McComb, author of Software Wasteland, quoted in Strategy + Business
See “Are you Spending Way too Much on Software at
https://www.strategy-business.com/article/Are-You-Spending-
Way-Too-Much-on-Software?

Call the model to reuse those rules whenever necessary
19
“You discover that many of the slight variations aren’t variations at all. They’re really the same things with
different names, different structures, or different labels. So it’s desirable to describe those few hundred concepts and
relationships in the form of a declarative model that small amounts of code refer to again and again.”
--Dave McComb (as previously cited)
See “Are you Spending Way too Much on Software at
https://www.strategy-business.com/article/Are-You-Spending-
Way-Too-Much-on-Software?

Deep transformation and a path to AI

What AI needs versus what it has
What it needs: Contextualized, disambiguated, highly relevant
and specific integrated data, flowing to the point of need
What it has: Single batch datasets cleaned up to be good enough
by data scientists, who spend 80% of their time on cleanup
What it needs: Knowledge engineers, and many bold Data
Visionaries in addition to big D Data Scientists, data-centric
architects, pipeline engineers, specialists in many new data niches
What it has: A growing group of tool users versed only in
probability theory, neural networks, python and R, including small
D data scientists, engineers and architects, plus scads of entrenched
application-centric developers
21
Finance
Operations
Marketing
Input Output
Input
layer
Hidden
layer 1
Hidden
layer 2
Output
layer

The real inhibitors to adoption aren’t technological – they’re rooted
in tribal biases and resistance to change
22
Tribalism CollectivismIndividualism
Anarchy TotalitarianismLocus of inertia
Daniel Quinn, Beyond Civilization and Alice Linsley, “Daniel Quinn: A Return to Tribalism?”, college-ethics.blogspot.com, 2018

Tribalism – Machine learning edition
23
Symbolists Bayesians Connectionists Evolutionaries Analogizers
Use symbols, rules, and
logic to represent
knowledge and draw
logical inference
Assess the likelihood of
occurrence for
probabilistic inference
Recognize and
generalize patterns
dynamically with matrices
of probabilistic, weighted
neurons
Generate variations and
then assess the fitness of
each for a given purpose
Optimize a function in
light of constraints
(“going as high as you
can while staying on
the road”)
Favored algorithm
Rules and decision trees
Favored algorithm
Naïve Bayes or Markov
Favored algorithm
Neural network
Favored algorithm
Genetic programs
Favored algorithm
Support vectors
Source: Pedro Domingos, The Master Algorithm, 2015
More at “Machine learning evolution”: http://usblogs.pwc.com/emerging-technology/machine-learning-evolution-infographic/, PwC, 2017

Tribalism – Data integration edition
24
Trend toward more data centricity this way
Application-centric
RESTful developers
Relational database
linkers
Data-centric
knowledge graphers
Application-centric
ESB advocates
Computerscience
wiki.org, 2018
TIBCO, 2014
Oracle DBA’s Guide, 2018
User Scott
Select FROM emp
Local
database
PUBLIC
SYNONYM
Emp - >
emp@HQ.ACME
>COM
Database link
(undirected)
Remote
database
EMP table
Portals
Net
Application
B2B
Interactions
Enterprise
data
Business
process
management
Web
Services
Mobile
Applications
DEE
Application
ERP
CRM SFA
Legacy
System
ESBCustom environment Common environment
API
Unstructured
Data
Semi-
structured Data
Structured
Data
Schema mapping based
on ontologies
Entity Extractor informs
all incoming data streams
about its semantics and
links them
Unified Views
RDF Graph
Database
PoolParty
Graph Search
Semantic Web Company, 2018

Crossing the chasm between the tribes
Reducing the amount of unfamiliarity developers confront--familiar document means to achieve
comparable ends to graph:
• Semantic suites that use the JSON format and familiar hierarchies: SWC’s PoolParty is an example
• GraphQL: A popular document shape language that talks to APIs using SELECT-like statements and tree shapes;
backend-agnostic; just uses a mental model for graph; addresses the API endpoint proliferation problem
• Accessible web as database methods: JSON-LD and Schema.org, etc. vocabularies
• Document “schemas” via data objects: JavaScript objects to developers = documents to NoSQL DB types;
Object Data Modeling instead of database semantics
• Mongoose or MongoDB JSON schema features + GraphQL: MongoDB object modeling and querying that
can be used for subdocument filtering within a GraphQL context
• HyperGraphQL: A GraphQL UI for Linked Data, restricted to certain tree-shaped queries
• Universal Schema Language: Mike Bowers’ document/graph query and modeling language still in development
• COMN: Ted Hills’ well-defined NoSQL + SQL data modeling notation
25

Transformative knowledge graphs
26PwC | Data-Centric architecture in business transformation

Types of logic most used in AI – Enabled systems
27
Rule-based systems (includes KR)
“Handcrafted knowledge” is the term DARPA
uses; rule-based programming + procedure
replication in process automation, + some
knowledge representation (KR)
• Strong on logical reasoning in specific
concrete contexts
- Procedural + declarative programming + set
theory, etc.
- Deterministic
• Can’t learn or abstract
• Still exceptionally common and useful
Statistical machine learning
• Probabilistic
• From Bayesian algorithms to neural nets
(yes, deep learning also)
• Strong on perceiving and learning (classifying,
predicting)
• Weak on abstracting and reasoning
• Quite powerful in the aggregate but
individually (instance by instance) unreliable
• Can require lots of data
Contextualized, model-driven approach
• Contextualized modeling approach—allows
efficiency, precision and certainty
• Combines power of deterministic,
probabilistic and description logic
• Allows explanations to be added to decisions
• Accelerates the training process with the help
of specific, contextual human input
• Takes less data
Example: Consumer tax software
Perceiving
Learning
Abstracting
Reasoning
Perceiving
Learning
Abstracting
Reasoning
Perceiving
Learning
Abstracting
Reasoning
Example: Facial recognition using
deep learning/neural nets
Example: Explains first how handwritten letters are
formed so machines can decide based on these
individual models—less data needed, more
transparency.
Previously dominant On the rise and rapidly improving Nascent, just beginning
John Launchbury of DARPA (https://www.youtube.com/watch?v=N2L8AqkEDLs), Estes Park Group and PwC research, 2017

Transformation scalability – The AirBnB knowledge
graph example
“In order to surface relevant context to people, we
need to have some way of representing
relationships between distinct but related entities
(think cities, activities, cuisines, etc.) on Airbnb to
easily access important and relevant information
about them….
These types of information will become increasingly
important as we move towards becoming an end-
to-end travel platform as opposed to just a place for
staying in homes. The knowledge graph is our
solution to this need, giving us the technical
scalability we need to power all of Airbnb’s verticals
and the flexibility to define abstract relationships.”
--Spencer Chang, AirBnB Engineering
28
Events
Neighborhoods
Tags
Restaurants
Users
Homes
Experiences
Places
Airbnb Engineering, 2018
Markets

Most automated knowledge graph – Diffbot?
“Diffbot’s crawler regularly refreshes the DKG with new information and its machine learning algorithms are smart
enough to pass over sites with histories of producing ‘logically inconsistent’ facts.
“‘That’s one of the reasons why we fuse information together from different sources,’ Tung said. ‘Our scale is such that
there’s minimal potential for errors. We’d bet the business on it.’
“Diffbot launched in 2008 and counts 28 employees among its core staff of engineers and data scientists.”
--Mike Tung of Diffbot, quoted in VentureBeat
Diffbot claims an automated knowledge graph of 1 trillion + facts, designed to grow without humans in the loop.
That compares with 1.6 billion crowdsourced facts in Google’s knowledge graph, according to VentureBeat.
29
Kyle Wiggers, “Diffbot launches AI-powered knowledge graph of 1 trillion facts about people, places, and things,”
VentureBeat, 30 August 2018

Versus more explicit, precise, contextualized meaning with a
triadic, Peircean knowledge graph and less than 1M concepts?
“There are many different approaches for distinguishing a logical basis for ontologies, but Peirce basically says to base
everything around 3s, explains [Mike Bergman of Cognonto]. That is,
1. the object itself;
2. what a particular agent perceives about the object;
3. and the way that agent needs to try to communicate what that is.
‘Without that triad it’s hard to ever get at differences of interpretation, context or meaning,’ he says, whether that be
between something like events and activities or individuals and classes.
Once you adopt that mindset, a lot of things that seemingly were irreconcilable differences begin to fall away, and the
categorization of information becomes really very easy and smooth....”
--Mike Bergman of Cognonto, quoted in Dataversity
30
Jennifer Zaino, “Cognonto Takes On Knowledge-Based Artificial Intelligence,” Dataversity, 23 November 2016

Contextual AI via a large knowledge graph at Fairhair.ai
31
Media Intelligence Apps
Global Monitoring Analyze & Report
Distribute Influence & Engage
New Apps
Employee
App
Freemium
At-Powered
Reporting
Outside
Insight
Enterprise
Solutions
Custom Solutions
3rd party Apps
PaaS
100M
documents ingested
daily
150 NLP/IR
pipelines
100’s Billions of
Searches
Service Layer
Context Building
Enriching & Analysis
Outside Data
Streaming, Search, Analytics, APIs
Building block to leverage the platform
Knowledge Graph
Enable cognitive applications on top of our data by connecting the dots
Data Enrichment Platform
Enrich, analyze & build by interoperating with all major players
AI-driven data Acquisition
Bring high quality outside to our repository with minimal human effort

Montefiore’s semantic data lake
32
HL7
feed
Web
services
EMR LIMS Legacy
OMICs CTMS
Claims
Annotation
engine
HDFS
Hadoop
HDFS
Hadoop
HDFS
Hadoop
HDFS
Hadoop
HDFS
Hadoop
HDFS
Hadoop
HDFS
Hadoop
HDFS
Hadoop
HDFS
Hadoop
HDFS
Hadoop
AllegrographAllegrographAllegrographAllegrograph Allegrograph
SDL loader
ML-LIB/R SPARQL
Prolog
Spark
Java API
Various data sources, some
structured, some not, now all
part of a knowledge graph
with a simple patient care-
centric ontology
Hadoop cluster with high-
performance processors
and memory
Scalable graph database
supporting open W3C
semantic standards
Standard open source
querying,ML and analytics
frameworks,
API accessibility
Doctors can query the graph
or harness ML + analytics and
receive answers from the
system at the point of care via
their handhelds.
The system also acts as a giant
feedback-response or learning
loop which learns
from the data collected via
user/system interactions.
Montefiore Health, Franz, Intel and PwC research, 2017

Siemens’ industrial knowledge graph
33
AI Algorithms
1 09:00 – Analyze
Turbine data hub
2 11:00 – Configure
Configure turbine
3 12:00 – Maintain
Master data Mgmt.
4 13:00 – Mitigate
Financial Risk Analysis
5 15:00 – Contact
Expert &
Communities
6 18:00 – Guide
Rules & Regulations
3
4
5
4
2
1
6
Industrial
Knowledge Graph
“Deep learning fails when it comes to context. Knowledge graphs can handle context
and enable us to address things that deep learning cannot address on its own.”
--Michael May, Head of Company Core Technology, Data Analysis and AI, Siemens

Pharma knowledge graphs for patient safety
Challenges
34
Solutions
Drug safety
Heightened
focus on safety
Evolving
regulatory
demands
Increasing
public scrutiny
Focus on
analytics
Increased
sharing &
transparency
Doing more
with the same
or less
Graph integration Natural language
processing
Data cleaning
during analysis
In-memory
query engine
PwC and Cambridge Semantics, 2018

NuMedii’s precision therapeutics knowledge graph
35
goTerm
Calcium ion
binding
2201
Protein
binding
Extracellular
region
ENSG
00000138829
Extracellular
matrix
disassembly
Extracellular
matrix
Organization
proteinaceous
extracellular
matrix
positive
regulation
of bone
mineralization
Fibrillin - 2
Extracellular
matrix
Structural
constituent
Extracellular
matrix
micro fibril
Camera-type
eye
development
CHEMBL_TC_
10038
Go Function
Reference_
Gosubset
prok
100001650
100001532
100001739
100000687
100002060
Ontotext and NuMedii, 2018

Thomson Reuters’ financial knowledge graph as a service
36
Thomson Reuters, 2018

Conclusion – Pitching to the boss
(in five minutes)

“What’s a knowledge graph?” Starting your 5-minute pitch
Minute 1: Define and describe the value proposition
of graphs.
• Graphs articulate relationship-rich data
• Compare graphs with tables: Relationships are what’s
missing from most data
• Describe relationships using graphs so machines can mine
them - essential to digitization
• Only way to build web-scale context in data is to
enrich relationships
• Without context, your organization can’t mature to
advanced analytics or real AI
• Document trees (e.g., taxonomies) are a stepping stone to
graph models
• Graphs are the parents that bring the data model family all
together, Tinker Toy-style
• Bottom line: Knowledge graph = a machine readable,
extensible model of your organization
• Build and maintain your advanced analytics/
AI foundation with that graph model
38
Graph:
Any-to-any
relationships
Document:
Nested, cumulative
hierarchies
Relational:
Row and column
headers and up-front
taxonomies
Relationship
richness
Relationship
sparseness
Static
Selective
Fragmented
Labor intensive
Additive
Index friendly
Immutable versioning
possible
More dynamic
More inclusive
More integrated
More machine assisted

“Who’s using knowledge graphs?” Only 9 out of 10 of the most
value-creating companies in the world
39
Company name Location Industry
Change in market
cap 2009-2018 ($bn)
Market cap
2018 ($bn)
1 Apple United States Technology 757 851
2 Amazon.Com United States Consumer Services 670 701
3 Alphabet United States Technology 609 719
4 Microsoft Corp United States Technology 540 703
5 Tencent Holdings China Technology 483 496
6 Facebook United States Technology 383(1) 464
7 Berkshire Hathaway United States Financial 358 492
8 Alibaba China Consumer Services 302(1) 470
9 JPMorgan Chase United States Financials 275 375
10 Bank of America United States Financials 263 307
v
Known
knowledge
graph builders
Operator of Taobao
and AliBot KG
builder
Known KG builders
Minute 2: List top users and explain why they’re using knowledge graphs.
• Apple using knowledge graph for Siri for years, doubling down on Siri knowledge engineers
• Alibaba's AliMe bot platform uses linked general and d0main-specific ontologies
• Amazon hiring knowledge engineers for Alexa ML training set acceleration
• Tencent one of the funders of Diffbot, an automated knowledge graph with 1 trillion facts
• Both Facebook and Google accelerate and enrich ML training set dev with their graphs
• B of A and Citibank both invested in the development of FIBO
(1) Change in market cap from IPO date
(2) Market cap at IPO date
Source: Bloomberg and PwC analysis

Innovation at scale can happen when machines learn from graphs
Minute 3: Machine learning on its own can’t build context. Top organizations understand that.
• Humans need to help machines identify and use context via the graph model of the organization
• Properly designed, the model can be extensible and scalable with humans and machines working together
• Contextual computing is the third wave of AI, according to DARPA—need to reposition for that wave
40

With a knowledge graph base, companies can skate to new
business models = Deep transformation
Minute 4:
• Once relationship data-enabled, organizations play
different roles than they've been accustomed to in the
digitized ecosystem.
• Some because of their data collection heritage can become
data providers.
• Others take up roles in the data supply chain, or position
themselves as industry platforms or marketplaces.
• Why are top companies able to cross industry
boundaries?
• Why can unicorns extend the reach of their
business models?
41

How transformative knowledge graphs work in practice –
Leading examples
Minute 5: Tell the stories of key companies and their knowledge graph journeys
42
Three deep transformation examples
AirBnB: Pursuing the goal of becoming an end-to-end
travel platform
Montefiore Health: Moving to a pay-for-performance business
model and a new level of profitability
Siemens: Scaling smart building with the help of
knowledge graphs
Three intelligent assistant examples
Alibaba: Extending the IA footprint of AliBot in general-purpose
customer service with graph models
Amazon: Boosting the performance of Alexa, particularly
improved interpretation and response
Apple: Acquired Lattice Data in 2017, investing in new Siri
knowledge engineering workforce hires in 2018
Companies like these are already skating to new growth and profitability. Data, many say, is the new oil. But most importantly,
relationship-rich data is the lubricant for your digital business engine.

Bonus points – Deep transformation via knowledge graph is also a
path to scalable data management and security
• Relationship data has long been overlooked, but specifying relationships is how you build context
• Connected, relationship-rich data will be seen as the most important asset for companies
• Can’t have governance without connected data, because disconnected data = sprawl
• Can’t have security if you can’t find or scale the unique identification and contextualization of your
data, so you can detect, track, trace, and know where it belongs
• Can’t have connected, meaningful data without a semantic model
• Can’t compete in the digital ecosystem and cross boundaries without meaningful data connectivity
• When it comes to enabling the scale and level of AI your company needs, including for data management and
security, knowledge graph + machine learning clears a path
43

Result – Graphs (including hybrids) complete the picture of your
transformed data lifecycle and how it’s managed
44

Questions or comments?
© 2018 PwC. All rights reserved. PwC refers to the US member firm or one of its subsidiaries or affiliates, and may sometimes
refer to the PwC network. Each member firm is a separate legal entity. Please see www.pwc.com/structure for further details.
pwc.com Alan Morrison
Sr. Research Fellow
PwC | Integrated Content | Emerging Tech
Mobile: +1(408) 205 5109
Email: alan.s.morrison@pwc.com
PricewaterhouseCoopers LLP
488 Almaden Blvd., Suite 1800
San Jose, CA 95110 USA
https://www.linkedin.com/in/alanmorrison
@AlanMorrison quora.com/profile/Alan-Morrison

Data-Centric Business Transformation Using Knowledge Graphs

Empfohlen

Empfohlen

Weitere ähnliche Inhalte

Was ist angesagt?

Was ist angesagt? (20)

Ähnlich wie Data-Centric Business Transformation Using Knowledge Graphs

Ähnlich wie Data-Centric Business Transformation Using Knowledge Graphs (20)

Mehr von Alan Morrison

Mehr von Alan Morrison (11)

Kürzlich hochgeladen

Kürzlich hochgeladen (20)

Data-Centric Business Transformation Using Knowledge Graphs