Data integration, data interoperation and data quality are major challenges that continue to haunt enterprises. Every enterprise either by choice or by chance has created massive silos of data in different formats, with duplications and quality issues. Knowledge graphs have proven to be a viable solution to address the integration and interoperation problem. Semantic technologies in particular provide an intelligent way of creating an abstract layer for the enterprise data model and mapping of siloed data to that model, allowing a smooth integration and a common view of the data. Technologies like OWL (Web Ontology Language) and RDF (Resource Description Framework) are the back bone of semantics for knowledge graph implementation. Enterprises use OWL to build an ontology model to create a common definition for concepts and how they are connected to each other in their specific domain. They then use RDF to create a triple format representation of their data by mapping it to the Ontology. This approach makes their data smart and machine understandable. But how can enterprises control and validate the quality of this mapped data? Furthermore, how can they use this one abstract representation of data to meet all their different business requirements? Different departments, different LoBs and different business branches all have their own data needs, creating a new challenge to be tackled by the enterprise. In this talk we will look at how the power of SHACL (SHAPES and Constraints Language), a W3C standard for defining constraint sets over data; complements the two core semantic technologies OWL and RDF. What are the similarities, the overlaps and the differences. We will talk about how SHACL gives enterprises the power to reuse, customize and validate their data for various scenarios, uses cases and business requirements; making the application of semantics even more practical.

1. ONE ONTOLOGY- ONE DATA SET- MULTIPLE SHAPES
SHACL
PRESENTED BY: Tara Raafat
October 2019
@ConnectedDataLondon

2. 2© Mphasis 2018 Proprietary and confidential information
Questions
Ø What is SHACL?
Ø How is it different from OWL ?
Ø How does it apply to the real world?

3. © Mphasis 2019 Proprietary and confidential information
A Quick Recap on RDF & OWL
• Simple triple based data model
• Subject, Predicate, Object
• Hitchcock is the director of “Psycho” movie.
• Individual things, and not just files, are given unique
identifiers.
• XML serialization (RDF/XML) for ease of data exchange
• Various textual representations for ease of human
understanding
• No Schema
Hitchcock Psycho
Is Director Of
http://www.filmarch.com/isDirectorOf
http://www.filmarch.com/Hitchcock http://www.filmarch.com/Psycho
• A knowledge model which defines a set of concepts and the
relationship between those concepts within a specific domain
• Supports automated reasoning and inference of data using logical
rules
• Provides Knowledge sharing and reuse among people or software
agents
RDF ONTOLOGY
OWL
3
• RDF based
• A defacto standard for ontology development
• Main components include
Classes : which define concepts in a domain
Properties: which are of two type
Object properties : define relationships between concepts
Datatype properties: define relationships between a
concept and a literal
Individuals: instances of classes
Restrictions: Allow definition of cardinality restrictions as well
existential and universal quantifications

4. © Mphasis 2019 Proprietary and confidential information
A Quick Recap on RDF & OWL
• Simple triple based data model
• Subject, Predicate, Object
• Hitchcock is the director of “Psycho” movie.
• Individual things, and not just files, are given unique
identifiers.
• XML serialization (RDF/XML) for ease of data exchange
• Various textual representations for ease of human
understanding
• No Schema
Hitchcock Psycho
Is Director Of
http://www.filmarch.com/isDirectorOf
http://www.filmarch.com/Hitchcock http://www.filmarch.com/Psycho
• A knowledge model which defines a set of concepts and the
relationship between those concepts within a specific domain
• Supports automated reasoning and inference of data using logical
rules
• Provides Knowledge sharing and reuse among people or software
agents
RDF ONTOLOGY
OWL
4
• RDF based
• A defacto standard for ontology development
• Main components include
Classes : which define concepts in a domain
Properties: which are of two type
Object properties : define relationships between concepts
Datatype properties: define relationships between a
concept and a literal
Individuals: instances of classes
Restrictions: Allow definition of cardinality restrictions as well
existential and universal quantifications
Director Film
Is Director Of
Producer Name
Hitchcock Psycho
Is A Is A

5. © Mphasis 2019 Proprietary and confidential information
OWL-STRENGHTS AND CHALLENGES
Ø OWL restrictions are not designed for validation- > they are there to allow inference of data.
Ø OWL is property oriented .
Ø OWL is for open world.
Ø Does not support defining specific validation constraints in different contexts.
Exactly 1 DoB
Exactly 1 name
Exactly 1 surname
Person
Who Is a person
Necessary &
Sufficient
:tara cus:name “Tara”;
Cus:surname “Raafat”;
Cus:DoB 10/10/2010 ;
RDFData
:tara a cus:Person
Infers
5

6. © Mphasis 2019 Proprietary and confidential information
Exactly 1 DoB
Exactly 1 name
Exactly 1 surname
A Simple ontology
Person
Customer
Account
Employee
Transaction
String
Risk Class
customerID
belongsTo
DateTime
customerSince
owns
in
performs
belongsTo Customer
Segment
Has Relationship Manager
String
employeeID
Branch
String
name
String
DateTime
String
surname
DoB
Height
String
Gender
Weight
Ethnicity
ID
EyeColor
String
ID
Ethnicity
String
String
SSN
worksIn
6

7. 3-Oct-19 | © Mphasis 2018 Proprietary and confidential information 7
SHACL

8. © Mphasis 2019 Proprietary and confidential information
SHACL
• RDF based
• Proposed as a W3C Recommendation in June 2017
• Close World Assumption
• Opposite logic than OWL. : for validation not inference
• Core elements
Shape
Target
Filter
Constraints
SHACL PROCESSOR
SHACL
Graph
Data Graph
Validation
Report
8
An RDF Validator

9. © Mphasis 2019 Proprietary and confidential information
SHAPES
SHAPES
Node
Shapes
Property
shapes
A conjunction of constraints that the targets must satisfy.
declare constraints directly on a node.
declare constraints on the property that
is connected to the node through a path.
:CustomerShape a sh:NodeShape;
sh:targetClass :Customer ;
sh:nodeKind sh:IRI .
Shape
:tara a :Customer . ✓
<http://banka.com/john> a :Customer . ✓
:7 a :Customer ✘
RDFData
:CustomerShape a sh:NodeShape;
sh:targetClass :Customer ;
sh:property[
sh:path cus:customerID ;
sh:dataType string;
sh:minCount 1;].
Shape
:tara a :Customer . ✓
cus:customerID “ER2198”.
:John a :Customer . ✘
cus:customerID
<http://banka.com/johnID>
RDFData
9
instance

10. © Mphasis 2019 Proprietary and confidential information
Paths
Ø There can be various path types:
Ø InversePath
Ø PredicatePath path: cus:name
Ø SequesncePath path: [cus:hasRelationshipManager cus:employeeID ]
Ø zeroOrMorepath
Ø alternativePath path: [ sh:alternativePath ( ex:father ex:mother ) ]
The sequence of edges through which a property is connected to a node
Customer
Employee
hasRelationshipManager
String
employeeIDmanagesRelationship
Person String
name
HighNetCustomer
10

11. © Mphasis 2019 Proprietary and confidential information
Target
Subject of
Property
Object of
Property Class(es)
Node(s)
Defines the focus of the constraints
:CustomerShape a sh:NodeShape;
sh:targetNode :tara :John ;
sh:nodeKind sh:IRI .
sh:property
[ sh:path cus:name ;
sh:minCount 1;
sh:maxCount 1;
sh:datatype xsd:string ; ] .
Shape
:CustomerShape a sh:NodeShape ;
sh:targetClass :Customer :Employee ;
sh:property [
sh:path cus:name ;
sh:datatype xsd:string . ]
Shape
CustomerShape a sh:NodeShape;
sh:targetSubjectOf cus: worksIn;
sh:property
[ sh:path cus:employeeID ;
sh:minCount 1;
sh:maxCount 1;
sh:datatype xsd:string ; ] .
ShapeShape
CustomerShape a sh:NodeShape;
sh:targetObjectOf cus: hasRelationshipManager;
sh:property
[ sh:path cus:employeeID ;
sh:minCount 1;
sh:maxCount 1;
sh:datatype xsd:string ; ] .
, rdfs:Class;
11

12. © Mphasis 2019 Proprietary and confidential information
☉ They are used to minimize the scope of the focus nodes
☉ Filters can be applied both on the target itself or at the
level of property
Filters
FilterTarget
:SeniorCustomer a sh:NodeShape ;
sh:targetClass cus:Customer;
sh:property [
sh:filterShape [
sh:property [
sh:predicate cus:age ;
sh:minInclusive 65 ; ]] ;
sh:predicate cus:accountType
sh:value :SeniorAccount ] ;
Shape
12
Customer
Tara
John
Jane
Mark
Stacy Tara
John
Jane
Mark
Stacy
Jane
Age>=65
Jane account
owns
account
Senior
Account
Rdf:type
Reduces the Focus nodes

13. © Mphasis 2019 Proprietary and confidential information
Property Constraints
Constraints
Ø Constraints are used for validation
Ø They are defined at the level of the node or at the level of the property
v Cardinality ( min , max)
v Class
v datatype
v nodeKind
v String
v Numeric
v Value
v hasValue
v in
Node Constraints
CustomerShape a sh:NodeShape;
sh:targetSubjectOf cust: worksIn;
sh:property
[ sh:path cust:employeeID ;
sh:minCount 1;
sh:maxCount 1;
sh:datatype xsd:string ; ] .
[ sh:path cust:owns ;
sh:Class :Account ; ]
Shape
:CustomerShape a sh:NodeShape;
sh:targetClass :Customer ;
sh:nodeKind sh:IRI .
Shape
13
Conditions that the targets must satisfy to be validated

14. © Mphasis 2019 Proprietary and confidential information
More interesting facts about SHACL
☉ Deactivate shapes -> sh:deactivated “true”
☉ Import shapes for re-usability
More things to do with SHACL:
☉ Business rule application
☉ interface building
☉ Data structure communication
☉ Code generation
☉ Data integration
14

15. 3-Oct-19 | © Mphasis 2018 Proprietary and confidential information 15
A real world
Scenario

16. © Mphasis 2019 Proprietary and confidential information
Bank use of Ontology and SHACL
Account
hasAmount 1M
Customer
HighNet
Customer
Transaction
Profile
Business Profile
hasProfile
subClassOf
OWL Model
OWL Restriction
RDF Data
:tara a :Customer .
cus:customerID “ER2198”;
cus:owns :Account231 .
:Account231 a cus:account
:Account231 cus:hasAmount 2M
:tara a :HIghNetCustomer .
Inferred Fact
:john a :Customer .
cus:customerID “JR17988”;
cus:owns :Account786 .
Cus:hasBusiness :bus786
:bus86 cus:businessType “retail”
:Account786 a cus:account
:Account786 cus:hasAmount 13k
:CustomerShape a sh:NodeShape;
sh:targetClass HIghNetCustomer;
sh:nodeKind sh:IRI .
sh:property
[ sh:path cust:hasProfile ;
sh:minCount 1;
sh:maxCount 1;
sh:class :TransactionProfile] .
CustomerShape a sh:NodeShape;
sh:targetNode cust: john;
sh:property
[ sh:path cust:performsTransaction” ;
sh:maxValue 10k;
] .
Ensuring every HNP customer
has a transaction Profile
CustomerShape a sh:NodeShape;
sh:targetSubjectOf cust:hasBusiness;
sh:property
[ sh:path cust:hasProfile ;
sh:minCount 1;
sh:maxCount 1;
sh:class :BusinessProfile ; ] .
Ensuring every customer who owns a
business has a business profile
Ensuring a certain customer does not
perform transactions over 10k
16

17. 3-Oct-19 | © Mphasis 2018 Proprietary and confidential information 17
Answers to the
questions
Ø What is SHACL and Why we need it?
Ø SHACL is an RDF Data Validator.
Ø How is it different from OWL ?
Ø OWL is for knowledge representation , knowledge sharing and
inference. SHACL is for data validation. OWL allows for data
integration and data interoperation whilst SHACL allows you to
have different views on that data (re-use it) and validate it
based on use cases.
Ø How does it apply to the real world?
SHACL
OWL
RDF
KNOWLEDGE
DATA
ECO-SYSTEM
conformsTo
ValidatedBy

18. overall_0_131831026671190437 columns_1_131831026671190437
THANK YOU
About Mphasis
Mphasis (BSE: 526299; NSE: MPHASIS) applies next-generation technology to help enterprises transform businesses globally. Customer centricity is foundational to Mphasis
and is reflected in the Mphasis’ Front2Back™ Transformation approach. Front2Back™ uses the exponential power of cloud and cognitive to provide hyper-personalized
(C=X2C2
TM=1) digital experience to clients and their end customers. Mphasis’ Service Transformation approach helps ‘shrink the core’ through the application of digital
technologies across legacy environments within an enterprise, enabling businesses to stay ahead in a changing world. Mphasis’ core reference architectures and tools, speed
and innovation with domain expertise and specialization are key to building strong relationships with marquee clients. Click here to know
Important Confidentiality Notice
This document is the property of, and is proprietary to Mphasis, and identified as “Confidential”. Those parties to whom it is distributed shall exercise the same degree of custody and care
afforded their own such information. It is not to be disclosed, in whole or in part to any third parties, without the express written authorization of Mphasis. It is not to be duplicated or used, in
whole or in part, for any purpose other than the evaluation of, and response to, Mphasis’ proposal or bid, or the performance and execution of a contract awarded to Mphasis. This document
will be returned to Mphasis upon request.
Presenter : Tara Raafat
Email: Tara.raafat01@Mphasis.com
Twitter:@taronish