The document discusses business ontologies and their potential as a transformative technology. It begins by introducing semantics and the progression from dictionaries to taxonomies to ontologies. Ontologies add logical assertions that define concepts and distinguish them from one another. The document then discusses several potential uses of ontologies, including technical applications like semantic querying and integration, as well as business uses such as a shared business vocabulary, regulatory reporting, and data refactoring. Overall, the document argues that ontologies can provide a common point of reference and shared meanings that transform how businesses work with data and systems.
Predictive Analysis - Using Insight-informed Data to Determine Factors Drivin...
Business ontologies
1. Business Ontologies:
A Transformative Technology?
Mike Bennett, Hypercube Ltd.
10th Annual Transformative Technologies Workshop
San Francisco Hilton
Friday, August 9, 2019
2. Outline
• Introducing Meaning
• Semantics: Dictionaries, Taxonomies, Ontologies
• A formal resource for semantics: business ontologies
• Uses for Ontology
• Technical applications
• Business uses
• Governance for corporate meaning
3. Demystifying Ontology
• Ontology is essentially about the meanings of things
• What is meaning (semantics)?
• NB forget the technology at this point!
5. What I say, what I mean
“Account”
source: http://www.businessdictionary.com/definition/account.html
Chronological record of changes in the value of an entity's assets, liabilities, and
the owners' equity; each of which is represented by a separate page in the ledger.
6. What I say, what I mean
“Account” “Ledger Account”
synonym
Profit and Loss Account
example
Chronological record of changes in the value of an entity's assets, liabilities, and
the owners' equity; each of which is represented by a separate page in the ledger.
7. What I say, what I mean
“Account” “Ledger Account”
synonym
A story of what happened
On-going contractual relationship between a buyer and seller
Profit and Loss Account
exampleThe Acme Account
example
source: http://www.businessdictionary.com/definition/account.html
Chronological record of changes in the value of an entity's assets, liabilities, and
the owners' equity; each of which is represented by a separate page in the ledger.
9. Logic and Semantics (Meaning)
Implication
• A simple set of logical statements
• implies a broad range of possible inferences (implications)
Implication Implication ImplicationImplication Implication
Some simple logic
10. Logic and Semantics (Meaning)
Implication Implication Implication ImplicationImplication Implication
• A broader set of logical statements
• Narrows down the range of possible inferences (implications)
• The more I say the less I (might) mean
More extensive logic
12. What is Meaning?
• When I say something I mean something
• I might mean exactly what I say
• I might mean more than I say
• That is:
• I might simply say what there is
• I might say things that imply more than I say
• I mean more than I say (implications)
• So we have:
• Basic statements: what there is
• Inferences: what else is implied
13. Meaning
• We have de-mystified ‘meaning’ or semantics
• Meaning is simply:
• What something refers to
• Logical implications of more complex statements or logic
• The meaning of what?
• A word
• A data element
• A complex sentence in natural language
• A set of logical statements
14. Transformative Technology
• Everything up to now is nothing to do with technology
• How do we put semantics to work in technology?
• How do we solve existing technical problems with semantics?
• Let’s start with a problem…
18. Dictionaries and Taxonomies
• Data dictionary?
• Business Glossary?
• Business dictionary
• Terminology?
• Thesaurus?
• Or what?...
• All these words about words…
18
20. Some History from the Finance Industry: The
FIBO Moment
• Previous standardization efforts at message and data levels
• Arguments over terms
• Mike Atkin: “What if we considered the concepts without worrying
about the words people use?”
• Sudden outbreak of peace!
• People could agree on the concepts just not on the ‘terms’ (words,
data element names)
20
23. LEXICAL SPACE
CONCEPTUAL SPACE
What Dictionaries Do
Word
Concept Concept Concept Concept Concept Concept Concept
Word Word Word Word Word Word
24. CONCEPTUAL SPACE
LEXICAL SPACE
What Dictionaries Do; Actually…
Word
Concept Concept Concept Concept Concept Concept Concept
Word Word Word Word Word Word
25. LEXICAL SPACE
CONCEPTUAL SPACE
Synonyms and Heteronyms (and other nyms)
Word
Concept Concept Concept Concept Concept Concept Concept
Word Word Word Word Word Word
• Some concepts have more than one word (synonyms)
• Some words mean more than one thing (heteronyms)
26. Dictionary Observations
• Each concept has a term (label) and a definition
• And some synonyms
• There is nothing to indicate these form a coherent set of concepts
• Some things might fall into two or more definitions
• Some definitions may be contradictory to other definitions
• Maintenance overhead
• Not auditable
• Yes, we use the term ‘audit’ for data and for semantics in much the same way
it is used in accounting or in IT Quality Assurance
28. CONCEPTUAL SPACE
LEXICAL SPACE
Dictionaries: Taxonomy
Word
Concept
Concept
Concept
Concept
Concept
Concept
Concept
Word Word Word Word Word Word
• We call this kind of structure a Taxonomy
Is a
Is a
Is aIs a
30. CONCEPTUAL SPACE
Ontology
Concept
Concept
Concept
Concept
Concept
Concept
Concept
• Ontology adds assertions about the things
Is a
Is a
Is aIs a
Concept Concept
Thing The set of things of which everything is a member
Is aIs a
Is aIs a
Is a
Concept
Concept
Concept
Concept
Concept
Verb (predicate)
Verb (predicate)
Verb (predicate)
Verb (predicate)
Verb (predicate)
More concepts
31. These relations are all describable in logic
• Predicate logic: ‘there is’
• Defines
• What exists
• What distinguishes each thing from another
• This is all understandable in basic set theory terms
• A kind of thing is a set the membership of which is defined in logical terms
• We need not concern ourselves with the notation or syntax for this
logic (just yet…)
32. Defining a Kind of Thing
Some kind
of thing
• We ask just two questions about this kind of thing:
• What kind of thing is it?
• What distinguishes it from other things?
32
33. What kind of thing is it?
Animal
Vertebrate Invertebrate
Bird Mammal Fish
Waterfowl
Some kind
of thing
33
34. What distinguishes it from other things?
Animal
Vertebrate Invertebrate
Bird Mammal Fish
Waterfowl
Some kind
of thing
Walks like a duck
Swims like a duck
Quacks like a duck
34
35. It’s a Duck!
Animal
Vertebrate Invertebrate
Bird Mammal Fish
Waterfowl
Walks like a duck
Swims like a duck
Quacks like a duck
35
36. The Two Ontological Questions
• For every kind of “Thing”:
• “What kind of Thing is this?”
• “What distinguishes it from other things?”
• These questions lead to:
• Taxonomic hierarchy of types of thing
• Properties of those things
• Simple facts: text, dates, numbers etc.
• Relationship facts: relate to other things
36
44. 44
This is not a more abstract
model of the solution…
Concept Model (CIM)
Logical Data Model (PIM)
Physical Data Model (PSM)
Realise
Implement
The Language Interface
Business
Technology
45. 45
This is not a more abstract
model of the solution…
Concept Model (CIM)
Logical Data Model (PIM)
Physical Data Model (PSM)
Realise
Implement
The Language Interface
Business
Technology
It’s a concrete model
of the problem!
46. Uses of Ontology
• Technical applications
• Semantic Web (reasoning / inference processing)
• Semantic queries
• Institutional knowledge graph
• Integration
• Governance:
• Single source of the concepts in the organization
• Sits above all of the above
• Resource owned and maintained by the business (stakeholders) not IT
48. Semantic Web Applications
• Application that uses logic to draw inferences from data
• These are a niche kind of application
• We call them ‘Puzzle solving’ applications
• These use a kind of application resource called an ‘ontology’
• The ontology is designed and optimized for the application
• It is designed around ‘competency questions’ (a kind of ‘use case’)
• It is focused on data
50. Inference Processing Example (FIBO)
50
Fixed_Float_IR_Swap_Different_Currencies_Single_Step_Notional_Contract
Fixed_Float_IR_Swap_Different_Currencies_Contract
Fixed_Float_IR_Swap_Contract
Interest_Rate_Based_Swap_Contract Currency_IR_Swap_Contract
Cross_Currency_IR_Swap_Contract
Faceted class
Interest_Rate_Swap_Contract
Rate_Based_Swap_Contract
Index Rate
Fixed_Rate_Leg
Currency
Single_Step_Notional_Amt
Party
Floating_Rate_Leg
Currency
Single_Step_Notional_Amt
Party
Swap_Contract Rate_Based_Derivatives_Contract
Interest_Rate_Derivatives_Contract
Swap
FIBO detects different currencies
Poly-
hierarchical
class
Poly-
hierarchical
class
Classification is based
upon the attributes of
the contract
Is both a Swap Contract and a Rate
Based Derivatives Contract
FIBO maps this to the
ISDA Fixed-Float-
Cross-Currency Swap
Derivatives_Contract
…
The Reasoner
performs semantic
reasoning to infer the
class of the swap
With acknowledgment to
David Newman, Wells Fargo
51. Ontology Technical Uses
• The inference processing (puzzle solving) application lets us draw
inferences from existing data
• These depend on both the data and the need for specific inferences to be
drawn (called ‘competency questions’)
• This is a Very niche application
• The ability of ontology to frame a ‘logical space’ that can map to
some set of things in the world (semantic space) provides many more
opportunities than simple puzzle solving
• Technical: semantic querying, reporting, decision support
• Business oversight of information assets
• Integration, development, QA
52. Virtual Semantic Data Lake: Reporting etc.
Risk, Compliance etc.
Semantic Queries
Legacy Data Sources and Systems
Ontology to Legacy Database Adapters
Reference Ontology
Reporting
53. Virtual Semantic Data Lake: Hybrid / Migration
Risk, Compliance etc.
Semantic Queries
Graph Triplestore
ETL
Query
Response
Inference
Processing
Legacy Data Sources and Systems
Ontology to Legacy Database Adapters
Reference Ontology
Reporting
55. Knowledge Graph: Data with Semantics
55
Visualization of instance data for a credit default swap, including data
types and classifications of many key attributes
With acknowledgment to
David Newman, Wells Fargo
56. Input to Graph Analytics Software
56
Source:
1. Semantic query
2. Query result interfaces with R
“igraph” package
Score = eigenvector centrality of
adjacent network positions
Node size reflects grand total aggregate
amount at risk for entity
Line width reflects aggregate amounts at
risk between individual trading parties
With acknowledgment to
David Newman, Wells Fargo
57. Ontology Business Uses
• Common point of reference across the business
• For humans
• For IT
• Reporting
• Management and risk dashboards
• Data refactoring
• E.g. client v entity data, reporting data
58. Aligning Business Glossary and Data Elements
Country A self-governing geopolitical unit
that is recognized as a country by
the United Nations
Country of Incorporation
This term refers to country in which a corporation is legally registered for operations.
This term is applicable to corporate entities only.
XYZ: Registered Address Country
Country of Birth
The country which customer documentation identifies as country of birth of the
customer. This term refers to individuals only.
PQR: Customer Birth place
Country of Residence
The country where the customer declared he / she resides . This attribute refers to
individuals only.
PQR: Customer Domicile Country Code
Domicile Country
This term refers to the country where the company operates for its primary operations.
This term is applicable to corporate entities only
XYZ: Customer Domicile Country Code
Country Code
This is a 2-3 digit standard used to refer to a country, where the code is applicable to
county of birth, county of residence, etc.
ISO 3166 Country Code
Country of Risk
This term refers to the country of risk for a customer's exposure with the Bank. This may
be related to the currency of transaction or to location of a transaction or payee. This is
applicable to corporations of individuals.
XYZ: Customer Risk Country Code
PQR: Customer Collateral Country Code
trace
Business concept context Data element context
ONTOLOGY
Contexts: Time Role / Relationship Records / history etc.
Business Glossary Data Model
Business Glossary Details
Analysis of a representative set of terms in a bank business glossary and accompanying data dictionary. Terms like ‘Country’ are defined in the context in which they are
used, for example country of domicile, risk etc. Meanwhile data elements need the context of class, class hierarchy in data models to determine semantics.
60. Regulatory Reporting Current State
60
FORMS FORMS
REPORTING ENTITY REGULATORY AUTHORITY
Reports (forms)
?
Uncertainty
• Content of the reports
• Are we comparing like with like?
• Data quality and provenance
Change in Reporting requirements =
• Redevelopment effort
• By each reporting entity
• For each system and form
61. Regulatory Reporting with Semantics
61
Thing
IR Swap CDS Bond
Contract
Common
ontology
Thing
IR Swap CDS Bond
Contract
Granular
data
REPORTING ENTITY REGULATORY AUTHORITY
Common
ontology
Data is mapped from each system of record into
a common ontology
Reported as standardized, granular data
Agnostic to changes in forms
Receives standardized, granular data aligned with
standard ontology (FIBO)
Uses semantic queries (SPARQL) to assemble
information
Changes to forms need not require re-
engineering by reporting entities
!
62. Ontology for Data Refactoring: Before
Legal Person
CRM1
Bank
Full name
Customer ID
Purchase date
Phone
Address Line 1
…
City
Country
CRMn
Name
Customer ID
Date of Birth
Purchase date
Phone
Address Line 1
…
Country of Collateral
CRM2
First name
Family name
Customer ID
Drawdown date
Credit rating
Address
Country of Birth
Data Siloes
There are various points at which the bank interacts
with a given entity or customer
63. Ontology for Data Refactoring: After
The Ontology makes implicit contexts explicit. It provides contextual semantics, distinguishing concepts specific to relationships, historical records (e.g. loan
applications), current information for entities, and others. This is independent of the deployment benefits of semantic technology.
CONCEPT (REFERENCE) ONTOLOGY
Current
date
Contexts: Role / Relationship Records / history etc.
Address
Entity Data
Customer ID
Purchase history
Contact Phone
Customer ID
Drawdown date
Payment history
Customer ID
Purchase date
Country of collateral
First name
Family name
Current Credit rating
Date of Birth
Home Phone
Country of Birth
Address Line 1
..
Address line n
City
State
Country
Full name
Customer ID
Purchase date
Application Credit rating
Phone
Address Line 1
…
City
Country of Domicile
Relationship
(context) specific data
Records: Loan
application data
Independent entity
data
Non context specific data
becomes real-time entity data
Ontology provides the context
for each kind of data
Purchase
date
Time
64. Managing the Concept Ontology
• Some say “Meaning is context”
• Correct:
• The meaning of words is determined by context
• The meaning of some data element or report content is very contextually
dependent
• The overarching reference ontology needs to be ‘above context’
• Then extract by context for
• Applications
• Reports
• Vocabulary
65. Concepts Extraction: Logical Models
N-dimensional content
Shown as 4D for simplicity
Context-specific models
Various models of content in
that hypercube in lower dimensionality
= Context
65
66. Concepts Extraction: Ontology
N-dimensional content
Shown as 4D for simplicity
Context-specific ontologies
Various extracts from
that hypercube in lower dimensionality
= Context
66
67. Vocabulary Layer by Context
N-dimensional content
Shown as 4D for simplicity
Context-specific concept models
Various extracts from
that hypercube in lower dimensionality
Vocabulary
Ontology
Local Context
Vocabulary
67
69. Lexical, Logical and Conceptual Spaces
LEXICAL SPACE
Word Word Word Word Word Word Word
70. Conclusions
• Semantics is simply what we mean by something
• What we mean by words
• What some data means
• Semantics can be
• Simple assertions – there is this and it is like that
• Implied knowledge – inferences
• Ontology as a business resource
• Owned and maintained by / for business
• Requires in house skills in taxonomy / concept theory
• Develop some simple techniques to deal with ‘the plumbing’ and bring in IT for that
• The plumbing:
• Inferencing applications
• Reporting / management dashboards
• IT Development governance
• Corporate governance
71. Thank you!
• Mike Bennett
• Director, Hypercube Ltd.
• www.hypercube.co.uk
• Email: mbennett@hypercube.co.uk
• Twitter: @MikeHypercube
A Member of the Semantic Shed