Wisdom of Enterprise

Knowledge Graphs
The path to collective intelligence
within your company
 Wisdom of Enterprise Knowledge Graphs | The path to collective intelligence within your company

“Intelligence is the ability to

adapt to change.”
Stephen Hawking

02
 Wisdom of Enterprise Knowledge Graphs | The path to collective intelligence within your company

Preface 04
Use Cases 06
Impact analysis 06
AI-based decision-making 07
Product DNA and root-cause analysis 08
Intelligent data governance 09
Context based knowledge transformation 10
Leverage knowledge from external sources 11
What are Knowledge Graphs? 12
The foundation for Knowledge Graphs 14
Conclusion 15
Contact 16

03
 Wisdom of Enterprise Knowledge Graphs | The path to collective intelligence within your company

Preface
Today, companies are relying more and more on artificial intelligence in
their decision-making processes. Semantic AI enables machines to solve
complex problems and explain to users how their recommendations
were derived. It is the key to collective intelligence within a company –
manifested in Knowledge Graphs.

Unleashing the power of knowledge is one That said, it is extremely time consuming
of the most crucial tasks for enterprises to share your domain knowledge. First you
striving to stay competitive. Knowledge, have to structure and adapt the informa-
however, is not just data thrown into a tion to fit into a pre-defined data model.
database. It is a complex, dynamic model
that puts every piece of information into a Knowledge Graphs connect knowledge
larger frame, builds a world around it and from different domains, data models and
reveals its connections and meaning in a heterogeneous data formats without
specific context. changing their initial form. They enable
industrial enterprises to harness the poten-
In reality,a database can only accommo- tial of collective intelligence.
date a small fraction of a company’s know-
ledge, much less the combined knowledge By design, an Enterprise Knowledge Graph
of a group of experts. A lot of vital informa- will easily connect all knowledge sources
tion is stored in unstructured documents, within a company, thereby enabling their
images and people’s minds. AI technologies to go beyond conventional
Machine Learning.
For decades, companies have been trying
to figure out an intelligent way to integrate
all expert knowledge in one magic place,
driven by a belief in the power concen-
The search for information
takes 14–30 percent of the
trated in the combined experience of
thousands of employees.

engineers’ time.

04
 Wisdom of Enterprise Knowledge Graphs | The path to collective intelligence within your company

Fig. 1 – Knowledge Graphs support highly complex decision-making by considering expert

knowledge from different domains. Real world dependencies and cross-correlations are
taken into account before recommendations are derived and explained to humans.

Which product variants

are affected by new EU
regulations?

Which parts produced by a

(sub-sub-) supplier will be in short
supply due to China's floods?

What happened
historically with price of
this sub-part when the
crude oil went up?

How could competitors disrupt

our supply chain by buying up
every producer of this chip?

05
 Wisdom of Enterprise Knowledge Graphs | The path to collective intelligence within your company

Use Cases

Impact analysis
The Enterprise Knowledge Graphs reveal the far-reaching
impacts of seemingly small decisions.

There are literally thousands of decisions imagine all of the possible future scenarios Example
being made across a company every arising from every single factor across an To reduce internal friction in a com-
day in areas like product modifications, enterprise in our intuitive risk calculations, bustion engine, the engineer reduces
production plans or the supply chain, but and we may fail to take key issues into the surface roughness by 10 μm
even small decisions can have huge con- account in our decision-making. without realizing that the n-tier sup-
sequences. Though science has effectively plier does not have the machinery
debunked the well-known butterfly effect, An Enterprise Knowledge Graph can assess required to produce this tolerance.
it is an undisputed fact that business deci- the far-reaching impacts of decisions The result: a massive investment and
sions can have a far-reaching impact. and their consequences for any area of a delayed start of production.
the company. It combines the human
Our ability to grasp the range of risks understanding of the real-world interde- Using a Knowledge Graph to connect
arising from a decision may, however, be pendencies with the power of computation. the engineering information with the
limited when we use an opportunistic As a result, the system can detect cross manufacturing information across
mindset. It is much too complex for us to correlations that humans cannot. the value chain alerts companies
to all possible risks arising from a
change of tolerance particularly in
Fig. 2 – The Butterfly Effect after a seemingly small decision: Knowledge Graphs terms of the supplier’s production
reveal the far-reaching consequences of a product change. capability. An AI-based analysis of
internal and external sources can
also assess the risk of sub-suppliers
going out of business and how that
would impact requirements. Even
though this is common practice for
OEMs – the only way to achieve truly
exhaustive results is to make the
connection with technical product
information.

06
 Wisdom of Enterprise Knowledge Graphs | The path to collective intelligence within your company

Use Cases

AI-based decision-making
Integration of all product data within a knowledge graph leads to process cost savings
of up to 65 percent and enables managers to make well-founded decisions.

Today’s companies are relying more and A Knowledge Graph is a new form of AI that Example
more on artificial intelligence in their allows us to answer the question "Why". It Complex mechatronic products such
decision-making process. Too often, these contains all of a company’s business logic as cars, airplanes and engineering
systems do not allow users to determine and allows to question, backtrack and machinery. usually have a large
why the AI recommends a certain course explain any recommendation proposed by number of variants. Each variant
of action – even though traceability can the AI. has its own set of configuration
be essential for critical cases like product rules. In cars, for example, a single
liability provided by the AI. component can impact 90% of the
other roughly 50,000 components on
average. It takes tens of thousands
Fig. 3 – A knowledge-based representation of real-world interdependencies of such configuration rules to map
enables machines to solve complex problems and explain to users how it these relationships.
derived its recommendations.
The components within the config-
urations are described by a limited
list of structured attributes. Human
intelligence can associate thousands
of partially unstructured pieces of
context information on each part in
terms of its function, geometry, logis-
tics, quality and many more areas.
This further increases the complexity
of the product structure.

A knowledge-based representation
of configuration rules allows us to
automate our analysis of the consist-
ency, completeness and redundancy
within the configuration rule set. The
system also automatically recom-
mends how to resolve a conflict. As it
is visualized in an intuitive graph, the
information is easily accessible for
human workers. Even unskilled users
can trace back any recommendation
to individual rules and conflicts.
Being able to enrich the configuration
knowledge with information about
logistics, finance and production ena-
bles AI powered systems to derive
even more complex recommenda-
tions, such as removing a configu-
ration from the product portfolio or
reorganising the assembly plan.

07
 Wisdom of Enterprise Knowledge Graphs | The path to collective intelligence within your company

Use Cases

Product DNA and

root-cause analysis
WIthin seconds, we can resolve complex product structures that are
configurable on many levels and compare different states.

The Enterprise Knowledge Graph not only tions and operations that caused them. We Example
connects information using enterprise can create – in milliseconds – a complete Where product failure is caused by a
business logic but also adds temporal con- DNA profile for any object that comprises combination of ambient conditions,
text. Thus, any part, service, money trans- all of the data associated with it throughout the batch of raw materials used or
fer or test result can be traced back in time its lifecycle, including the resolution of the the supplier production site, a classic
through the different versions, transforma- entire production structure. root-cause analysis can be very time
consuming and may only produce
limited results.
Fig. 4 – Semantic root-cause analysis based on a Knowledge Graph.
Knowledge Graphs can statistically
analyze root causes by connecting
data from different domains without
creating a hypothesis in advance. The
statistical analysis uses the intelli-
gence within the Knowledge Graph to
avoid spurious correlations.

08
 Wisdom of Enterprise Knowledge Graphs | The path to collective intelligence within your company

Use Cases

Intelligent data
governance
Knowledge Graphs store the entire business logic of an enterprise and
enable intelligent master data governance.

Given the changing market conditions, reasoning. Merging data models from dif- Example
digitalization of product development, ferent sources and creating rules to avoid The system landscape at an auto-
and increasing product complexity and inconsistencies will become an intuitive and motive OEM has evolved over time
customization, companies are faced with efficient process. and presents data inconsistencies
new challenges that require advanced data between different systems. Materials
interaction. The semantic backbone relies on cus- with the same UUIDs have different
tomized metalevel ontologies based on attributes, and materials with same
Knowledge Graphs enable semantic tech- the graphs, which help to connect data attributes often have different identi-
nologies to automate processes like data from different systems and guarantee fiers. Integrating the sources within a
structuring, text analysis and data model consistency through logical inferencing and semantic backbone based on a
merging. They can also integrate structured similarity calculations. At the same time, Knowledge Graph enables the com-
and unstructured data and facilitate the machine learning component derives pany to monitor data quality very
data clean up and enrichment based on rules to automatically detect redundancies quickly using semantic similarity.
and inconsistencies in the data. Redundant, inconsistent and incom-
plete data is quickly identified and
Fig. 5 – Connecting data-sources by storing the transformation logic automatically corrected. At the same
within a Knowledge Graph. time, the machine learning compo-
nent derives rules based on input
data and cleans up data even as it
Database A improves the data correction process.

Material ID 78543
Revision 01
…

Database B

Search-Nr. 78543-01
…
…

Scattered data relates

Database C to the same concept in
the Knowledge Graph
Product ID 7854301
…
…

09
 Wisdom of Enterprise Knowledge Graphs | The path to collective intelligence within your company

Use Cases

Context-based knowledge
transformation
Knowledge Graphs automatically transform and represent knowledge in the
required context for any user.

Information can only evolve into knowledge cial data, engineering data, manufacturing Example
by adding context to it. The act of simply data and procurement data. Depending A very common example is the dif-
retrieving information on its own is basi- on their tasks, users access the same data ference between how engineers and
cally useless without a deep understanding with a different point of view that serves manufacturing teams perceive the
of the context, which may vary strongly their respective purpose. Storing product product structure. A Bill-Of-Materials
depending on what the information will structure data and transformation infor- (BOM) is a structured list of product
ultimately be used for. A mechanical part is mation within a Knowledge Graph allows assemblies, subassemblies, parts,
associated with, among other things, finan- the user to explore the database within the etc. included in the product. The
required context. BOM can have different levels of
maturity depending on when it is
Fig. 6 – Context based views on data within the Knowledge Graph. used during the product lifecycle
(development, production, after-
The Knowledge Graph can be regarded as a cloud sales/maintenance). The engineering
of data points watched form different perspectives BOM (E-BOM) contains the hier-
und thus presenting the same object in different contexts. archical structure of a product as
designed by the engineer. However,
when drafting the E-BOM, engineers
don’t factor in how and where the
product will be manufactured.
Depending on the component inter-
faces specified in the CAD model, we
can create a site-specific manufactur-
ing BOM (M-BOM) for specific manu-
facturing processes at the production
site or the suppliers. This may entail
replacing certain materials, restruc-
turing the production process based
on physical limitations or adding
As maintained non-design items like oil or glue that
are not mentioned in the E-BOM.
As designed

As built

10
 Wisdom of Enterprise Knowledge Graphs | The path to collective intelligence within your company

Use Cases

Leverage knowledge from

external sources
Knowledge Graphs easily integrate data from external databases, deep web, open
sources and geographic information systems and save costs caused by regulatory
and contract conflicts as well as unforeseen risks in companies' ecosystem.

Managing data generated within the Knowledge Graphs are continuously Example
company becomes a lot more difficult enriched with information from external Whether or not a product complies
when you have to consider external factors regulatory databases, news articles, mate- with prevailing regulations and con-
e.g. legal regulations. The semantic core rial databases, supplier information, geo- tractual terms will depend on a range
of Knowledge Graphs makes it easy to spatial data etc. The data is automatically of different factors including product
integrate external ontologies, vocabularies processed with intelligent NLP algorithms functions, production process param-
and databases. and integrated into the graph in a struc- eters, storage and transport as well
tured and understandable way. as the respective geographic regions.
This information may vary over time
and may be stored in part in an
Fig. 7 – Automated integration of unstructured information relating to external unstructured or heterogeneous form,
regulations or internal contracts and detection of potential compliance issues. which will require manual analysis.

Contracts External Product and process complexity

with clients regulations can be represented in a Knowledge
Graph, enabling users to apply
AI-based technologies for analytics
and data management such as NLP,
text clustering and semantic classi-
fication. The system can automate
regulatory compliance checks and
offer recommendations to resolve
compliance conflicts. It can also eval-
uate the similarities of different reg-
ulations and contracts. This will allow
Structuring and examination risk management to easily simulate
of process regarding different scenarios based on poten-
fulfillment of external and tial regulatory changes and analyze
internal restrictions. how they impact the company in
order to validate business decisions.
Similarly, Knowledge Graphs can be
used for homologation purposes in
the automotive sector.

11
 Wisdom of Enterprise Knowledge Graphs | The path to collective intelligence within your company

What are Knowledge Graphs?

A Knowledge Graph is a close-to-reality model that represents a
company’s business logic and serves as its central knowledge platform.
You could say it is the brain of a company, as its architecture is very
similar to that of our brains – unlike traditional databases and data
lakes. Both connect objects directly with each other rather than storing
data in different tables and then connecting them via JOIN-tables.

Experiment:
Think of something blue! Blue
You probably thought of the sky, a
car, a shirt or some other object with
the color blue that you have recently
seen. You brain directly associates all
of these things with the named color.
It behaves like a graph, a network of
entities where the entity “blue” clas-
sified as a color is directly connected
to cars and shirts that are blue. If
our brain worked the same way a
traditional relational database does,
it would have to look at every item in
the entire closet and check the value
of the attribute “color”. Obviously,
that would be a lot more time con-
suming – both for our brains and for
an enterprise database.

A Knowledge Graph combines sophisti-

cated AI and graph technologies. It is a
complex network that accesses data from
databases, silos and unstructured sources,
bringing it into context. A Knowledge Graph
connects different domain knowledge
models without altering them and evolves
over time with no need to resstructure
the models. A Knowledge Graph is able to
store complex models and interdepend-
encies that exist in the real world (process
instructions of a chemical productions
process, configuration rules of a complex
mechatronic product, etc.).

12
 Wisdom of Enterprise Knowledge Graphs | The path to collective intelligence within your company

Knowledge Graphs
connect, structure and
simplify knowledge within
organizations, creating
transparency by bridging
data silos.

13
 Wisdom of Enterprise Knowledge Graphs | The path to collective intelligence within your company

The foundation for

Knowledge Graphs
The creation and management of Knowledge Graphs requires
3 components: capturing the business logic and intelligence
of domain experts, connection of domain-specific business
logics and their representation in a machine-readable form and
storage of the knowledge graph in a semantic graph database.

Component 1: Semantics Cpmponent 2: Ontology Compnent 3: Graph Database

capturing Intelligence by connecting Methodology of representing intelli- Technology to store and manage the
information. gence in a machine-readable way. intelligence.
Semantic description is the process by An ontology is a formal representation of Graph databases enable storage, manage-
which we enrich data with machine- knowledge within a particular domain that is ment and analysis of highly connected data
readable meaning by creating the context created by defining its terms and concepts made up of objects and the relationships
for an object. Rules and relationships and enriching them semantically – a between them (nodes and edges). Graph
between data define its context. specification of a conceptualization. databases store Information in directly
connected objects instead of tables and
thus do not require resource-intensive
JOIN-operations.

Expert C

Expert A
Expert B

AI

14
 Wisdom of Enterprise Knowledge Graphs | The path to collective intelligence within your company

Conclusion
Knowledge is every company’s most Knowledge Graphs integrate data across
valuable asset, but it remains hard to grasp the whole enterprise, support complex
at the same time, because it is scattered decisions and reveal the origin of every
across different systems and human minds. correlation within milliseconds. It forms
The key to integrating knowledge efficiently the foundation for AI technologies that are
among systems and human users is to pro- more intelligent than artificial.
vide representation in machine-readable
form.

Creating a Knowledge Graph with seman- The Enterprise Knowledge

tic description of information context
allows users to access a machine-readable
representation of complex interdepend-
Graph is a close-to-reality
data model that contains the
encies that form a real-world model of the
knowledge domain.

business logic of an enterprise

and is stored in a graph
database.

15
 Wisdom of Enterprise Knowledge Graphs | The path to collective intelligence within your company

Contact

Philipp Obenland Ulrich Schoof

Director | Strategy & Operations Senior Manager | Strategy & Operations
Phone: +49 (0)89 29036 7822 Phone: +49 (0)151 5807 0972
[email protected] [email protected]

Boris Shalumov
Senior Consultant | Strategy & Operations
Phone: +49 (0)151 5807 4474
[email protected]

16
 Wisdom of Enterprise Knowledge Graphs | The path to collective intelligence within your company

Live Enterprise – a paradigm change

for companies

Dynamic leadership based on real time data pools, which enable continuous eval-
data: this approach realizes the potential uations in near real time with innovative
of new digital technologies. analytical methods and Machine Learning.
The time lag between analyzing the cur-
Welcome in the Here and Now! Digitaliza- rent state and implementing it into con-
tion makes this possible. Today more and crete action shrinks potentially towards
especially more actual data are available zero. The result: Relevant data, adaptive
than ever before. Thus, companies are processes, agile decision making –
gaining more precise fundamentals for a new form of leadership. Live Enterprise
their actions. is more than just an IT project.

They become a Live Enterprise: The This publication is one of our building
Internet of Things is the new data supplier blocks how to structure and connect data
and allows the generation of consistent on the journey to a live enterprise.

17
 Wisdom of Enterprise Knowledge Graphs | The path to collective intelligence within your company

This communication contains general information only not suitable for

addressing the particular circumstances of any individual case and is not
intended to be used as a basis for commercial decisions or decisions of any
other kind. None of Deloitte Consulting GmbH or Deloitte Touche Tohmatsu
Limited, its member firms, or their related entities (collectively, the “Deloitte
network”) is, by means of this communication, rendering professional advice or
services. No entity in the Deloitte network shall be responsible for any loss
whatsoever sustained by any person who relies on this communication.

Deloitte refers to one or more of Deloitte Touche Tohmatsu Limited, a UK

private company limited by guarantee (“DTTL”), its network of member firms,
and their related entities. DTTL and each of its member firms are legally
separate and independent entities. DTTL (also referred to as “Deloitte Global”)
does not provide services to clients. Please see www.deloitte.com/de/
UeberUns for a more detailed description of DTTL and its member firms.

Deloitte provides audit, risk advisory, tax, financial advisory and consulting
services to public and private clients spanning multiple industries; legal advisory
services in Germany are provided by Deloitte Legal. With a globally connected
network of member firms in more than 150 countries, Deloitte brings world-class
capabilities and high-quality service to clients, delivering the insights they need to
address their most complex business challenges. Deloitte’s approximately
286,000 professionals are committed to making an impact that matters.

Issue 07/2019
18