Build Knowledge Graphs with Oracle RDF to Extract More Value from Your Data

Analytics and Data Summit 2019
Build Knowledge Graphs with Oracle RDF to Extract More
Value from Your Data
Souri Das Matthew Perry Melliyal Annamalai
Oracle Spatial and Graph
Spatial and Graph Summit @

Copyright © 2019, Oracle and/or its affiliates. All rights reserved. |
Safe Harbor Statement
The following is intended to outline our general product direction. It is intended for
information purposes only, and may not be incorporated into any contract. It is not a
commitment to deliver any material, code, or functionality, and should not be relied upon
in making purchasing decisions. The development, release, timing, and pricing of any
features or functionality described for Oracle’s products may change and remains at the
sole discretion of Oracle Corporation.
Confidential – Oracle Internal/Restricted/Highly Restricted 2

Agenda
3
Definition and Use Cases for Knowledge Graphs
Modeling Knowledge Graphs using W3C RDF
Using Oracle RDF Knowledge Graph
Performing Analytics with RDF Knowledge Graph Data
1
2
3
4

Agenda
4
1
2
3
4

What is a Knowledge Graph?
• Knowledge graphs are large networks of entities, their semantic types,
properties, and relationships between entities1.
• Key Features:
– Things not strings
• global unique identifiers
– Formal structure/semantics
• machine processable, unambiguous
– Linked descriptions
• resources are described by their connections
1. M. Kroetsch and G. Weikum. Journal of Web Semantics: Special Issue on Knowledge Graphs.
http://www.websemanticsjournal.org/index.php/ps/announcement/view/19 [August, 2016]
Database
Knowledge
Base
Graph
Knowledge
Graph

Knowledge Graph Examples1
1. Heiko Paulheim. Journal of Web Semantics: Special Issue on Knowledge Graph Refinement. Preprint.
“Knowledge Graph Refinement: A Survey of Approaches and Evaluation Methods.” [September, 2016]

Use Case: Linked Data Publishing
• Publishing data in a standard way
so that it can be more easily
consumed
• Popular way for government
agencies to publish public data
• Why RDF?
– URIs
– Flexible data model
– Standard vocabularies
– Standard protocols
Italy
National Institute of Statistics
Japan
National Statistics Center

Use Case: Linked Data Publishing
• Publishing data in a standard way
so that it can be more easily
consumed
• Popular way for government
agencies to publish public data
• Why RDF?
– URIs
– Flexible data model
– Standard vocabularies
– Standard protocols
https://lod-cloud.net/ 1,200+ linked datasets

Business Challenge
• Link database information on genes,
proteins, metabolic pathways, compounds,
ligands, etc. to original sources.
• Increase productivity for accessing, sharing,
searching, navigating, cross-linking,
analyzing internal /external data
Solution
• Semantic integration layer on RDF graph
• Rich domain-specific terminology (biology,
chemistry and medicine) 1.6 M terms
• Terminology Hub: 8 GB of referential data
that cross-references between data
repositories.
Use Case: Semantic Data Integration
Novartis Institutes for BioMedical Research (NIBR)

Use Case: Smart Engineering/Construction
PDF
PDF
PDF
PDF
Build
Order
Engineering
Standards
Government
Regulations
Product
Specifications
Requirements and Specifications
Ontology
Manual assessment by
discipline-specific
engineers
Automated reasoning
and queries
Example: bolt length
error rate 15%  0.5%

Agenda
11
1
2
3
4

Copyright © 2019, Oracle and/or its affiliates. All rights reserved. | 12
W3C Standards for Knowledge Graphs
The World Wide Web
Consortium has defined a suite
of standards to support Linked
Data and Knowledge Graphs.
Fundamental Concepts are:
• URIs
• Links to other resources
• Standard Data Model (RDF)
• Standard Ontology Language
(OWL)
• Standard Query (SPARQL)

What is Resource Description Framework (RDF)
• An RDF graph is a directed, labeled
graph with some syntactic restrictions
– Vertex labels are URIs or scalar values
– Edge labels are URIs
– Source vertex for an edge must be a URI
• An edge is the atomic unit of an RDF
graph – called an RDF triple
univ:university1
univ:course456
univ:professorXYZ
univ:student123
univ:employedBy
univ:offers
univ:takes
univ:teaches
"John"
univ:name
"Susan"
univ:name
univ:university1 "MIT"
univ:name
Subject Predicate Object
RDF Triple
@prefix univ: <http://univ.org#>
univ:student123 == <http://univ.org#student123>
URI prefix and prefixed name

What is Resource Description Framework (RDF)
• Resources can belong to types or
Classes
• Distinguish between Classes and
Instances
– Classes:
• univ:Professor, univ:University
– Instances:
• univ:professorXYZ, univ:university1
univ:university1
univ:course456
univ:professorXYZ
univ:student123
univ:employedBy
univ:offers
univ:takes
univ:teaches
"John"
univ:name
"Susan"
univ:name
univ:Professor
univ:University
rdf:type
rdf:type

Example: Creating an RDF Knowledge Graph

Example: Creating an RDF KG for the University Domain
@prefix univ: <http://univ.org#> .
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
@prefix owl: <http://www.w3.org/2002/07/owl#> .
# Person
univ:Person rdf:type owl:Class .
# Student
univ:Student
rdf:type owl:Class ;
rdfs:subClassOf univ:Person .
# Graduate Student
univ:GraduateStudent
rdfs:subClassOf univ:Student .
# Professor
univ:Professor
rdfs:subClassOf univ:Person .
RDF Serialization (Turtle Format)
Defining Classes (and relationships among classes)
Person
Student Professor
Graduate
Student

# Course
univ:Course rdf:type owl:Class .
# Graduate Course
univ:GraduateCourse
rdfs:subClassOf unv:Course .
# University
univ:University rdf:type owl:Class .
Defining Classes (and relationships among classes)
Person
Student Professor
Course
University
Graduate
Course
Graduate
Student

# takes
univ:takes
rdf:type owl:ObjectProperty ;
rdfs:domain univ:Student ;
rdfs:range univ:Course .
# teaches
univ:teaches
rdfs:domain univ:Professor ;
# offers
univ:offers
rdfs:domain univ:University ;
Defining Properties
Person
Student Professor
Course
University
takes teaches
offers
Graduate
Course
Graduate
Student

# Graduate Students are Students who take a Graduate Course
owl:GraduateStudent owl:equivalentClass
[ rdf:type owl:Restriction ;
owl:onProperty univ:takes ;
owl:someValuesFrom univ:GraduateCourse
] .
Adding More Semantics
Person
Student Professor
Course
University
takes teaches
offers
Graduate
Course
Graduate
Student
takes

Person
Student Professor
# Student 123 takes Graduate Course 456
univ:student123 univ:takes univ:course456 .
univ:course456 rdf:type univ:GraduateCourse .
# what can we infer?Course
University
takes teaches
offers
Graduate
Course
Graduate
Student
Creating Edges (at instance-level)
Student123
Course456
takes

Person
Student Professor
# what can we infer?
# domain of takes is Student
univ:student123 rdf:type univ:Student .
Course
University
takes teaches
offers
Graduate
Course
Graduate
Student
Inference (at instance-level)
Student123
Course456
takes

Person
Student Professor
# Student is a subclass of Person
univ:student123 rdf:type univ:Person .
Course
University
takes teaches
offers
Graduate
Course
Graduate
Student
Student123
Course456
takes

Person
Student Professor
# student123 takes a Graduate Course
univ:student123 rdf:type univ:GraduateStudent .
Course
University
takes teaches
offers
Graduate
Course
Graduate
Student
Student123
Course456
takes

Person
Student Professor
# student123 takes a Graduate Course
univ:student123 rdf:type univ:GraduateStudent .
# range of takes is Course and GraduateCourse subclass of Course
univ:course456 rdf:type univ:Course .
Course
University
takes teaches
offers
Graduate
Course
Graduate
Student
Student123
Course456
takes

Graphical Tools like Protégé Make Things Easier

Agenda
26
1
2
3
4

Database Version and Options for
RDF Knowledge Graph
• On-premise
– Oracle Database Enterprise Edition with
Spatial & Graph and Partitioning options
• Oracle Database Cloud Services
– DBCS (OCI)
– Exadata Cloud Service
– ATP-Dedicated (planned)
27
https://blogs.oracle.com/oraclespatial has a 3-part series
on getting started with RDF on DBCS
“Using RDF Knowledge Graphs in the Oracle Public Cloud”

Oracle Spatial and Graph 18c – RDF Knowledge Graph Architecture
Generic Relational Schema
for Storing RDF Data
RDF Views of Relational Data
SPARQL-to-SQL
Query Translator
Forward-chaining
OWL Reasoner
RDF Bulk Loader SPARQL Update
Processor
SQL and PL/SQL API
Support for Apache Jena (Java API)
Protégé Plugin Fuseki Endpoint Cytoscape Plugin
SQL Developer
RDF Support
Enterprise Manager
and Other DB Tools

Loading RDF Data

Source for RDF data
• Formatted File
– RDF formatted files: N-Triple, N-Quad, Turtle, RDF/XML, Trig, JSON-LD, etc.
• SPARQL Update
– Incremental insert, add, copy, etc.
• (External) Tables/Views or SQL Query
– Any tabular rowsource + W3C RDB2RDF mapping
Formatted File or String, Tabular Rowsource
30

RDF Bulk Loader
31
Staging Table
Client DB Server
RDF Relational Schema
Parallel RDF Bulk Load
• Value Checking
• Canonicalization
• Duplicate Elimination
• Internal ID Generation
• Index Creation
Step 2
N-Triple N-Quad
N-Triple
N-Quad
Turtle
RDF/XML
Trig
JSON-LD
Others
External Table
Oracle
Protégé
Plugin
Jena Adapter
Java API
Step 1
Oracle
SQL*Loader

Relational Data to RDF (W3C RDB2RDF)
• Direct Mapping
– Automatically generates a mapping based on an input relational schema.
– A Direct Mapping of Relational Data to RDF.
W3C Recommendation 27 September 2012
http://www.w3.org/TR/2012/REC-rdb-direct-mapping-20120927/
• R2RML (RDB to RDF Mapping Language)
– An RDF vocabulary for specifying customized relational data to RDF mapping.
– R2RML: RDB to RDF Mapping Language.
W3C Recommendation 27 September 2012.
http://www.w3.org/TR/2012/REC-r2rml-20120927/
Two types of mapping: Direct and R2RML
32

Using Direct Mapping: Overall Flow
Direct Map
Author
RDB2RDF
Processor
SPARQL to SQL
Translator
List of Tables
and Views
R2RML map
& schema
33
RDF
Triplestore
Relational
Tables & Views
Query
Writer
SPARQL
QUERY

Using R2RML: Overall Flow
R2RML Map
Author
RDB2RDF
Processor
SPARQL to SQL
Translator
R2RML
Mapping
R2RML map
& schema
34
RDF
Triplestore
Relational
Tables & Views
Query
Writer
SPARQL
QUERY

R2RML Schema (Classes and Relations)
Source: Das, Sundara, and Cyganiak.
R2RML: RDB to RDF Mapping Language. W3C Recommendation 27 September 2012.
http://www.w3.org/TR/2012/REC-r2rml-20120927/

Example: Mapping EMP and DEPT Tables to RDF
7782 John DB 10
constraints
EMP
ENO ENAME EXPERTISE DNO
10 Sales NYC
DNO DNAME LOC
DEPT
pkey ref. pkey
x:Department
http://x.com/Dept#{DNO}
class
Subject
Predicate-object pairs
<http://x.com/Dept#10>
rdf:type x:Department ;
<../Dept/Deptno> 10 ;
<../Dept/DeptName> “Sales” ;
<../Dept/Location> “NYC” .
TriplesMap
 DEPT

Example: Mapping EMP and DEPT Tables to RDF
7782 John DB 10
constraints
EMP
ENO ENAME EXPERTISE DNO
10 Sales NYC
DNO DNAME LOC
DEPT
pkey ref. pkey
x:Employee
http://x.com/Emp#{ENO}
Predicate-object pairs
Subject
class
<http://x.com/Emp#7782>
rdf:type x:Employee ;
<../Emp/Empno> 7782 ;
<../Emp/EmpName> “John” ;
<../Emp/Expertise> “DB” ;
<../Emp/DeptNum> 10 ;
<../Emp/Department> <http://x.com/Dept#10> .
TriplesMap
 EMP

Oracle SQL Developer RDF Support
40
• R2RML Mapping Editor
– Shows a tree view of an
R2RML mapping
– Allows drill down from
Triples Map -> Subject
Map -> Predicate-Object
Map
– You can edit or delete
existing elements or add
new ones
– “Commit Mapping”
refreshes the associated
RDF view

Performing Inference

Core Inference Features in Oracle Database
42
• Oracle provides native inference in the database for
• RDFS, RDFS++
• OWLPRIME, OWL2RL, OWL2EL, SKOSCORE
• User-defined rules
• User-defined inferencing
• Inference done using forward chaining
• Triples inferred and stored ahead of query time
• Removes on-the-fly reasoning and results in fast query times
• Proof generation
• Shows one deduction path

• Option 1: Add user-defined rules
• Example: sibling rule
• Option 2: Leverage external DL reasoners
• Option 3: User-defined inferencing in Oracle Database Release 12c
Extending Semantics Supported by Native OWL Inference Engine
Antecedents

Consequents
?z :parentOf ?x .
?z :parentOf ?y .
?x owl:differentFrom ?y .
?x :siblingOf ?y
43

Extensible Architecture for External OWL DL Reasoners
External
In-Memory OWL
DL Reasoners
TrOWL/REL
Jena APIs
through
Adapter for
Apache Jena
Oracle’s
Native Inference
Engine for OWL 2
RL, EL & user-
defined rules
Materialized Inference
44

• SDO_SEM_INFERENCE.
INF_EXT_ACTION_START
INF_EXT_ACTION_RUN
INF_EXT_ACTION_END
User-Defined Inferencing in Oracle Database Release 12c
45

SPARQL Query and Update

What is SPARQL?
• SPARQL Protocol and RDF Query Language
– W3C standard for querying and manipulating RDF content
– Queries/updates and corresponding results are communicated via HTTP with a
SPARQL endpoint
– A SPARQL endpoint implements the SPARQL protocol and serves RDF data from a
RDF triplestore or RDF view

SPARQL Graph Pattern
Basic unit of SPARQL queries
univ:Student
univ:student123
univ:student456
"John Green"
"1999-06-15"^^xsd:date
"male"
"Susan Blue"
"2000-02-10"^^xsd:date
"female"
rdf:type
rdf:type
foaf:name
vcard:BDAY
foaf:gender
foaf:name
vcard:BDAY
foaf:gender
univ:Student
univ:student123
"John Green"
"1999-06-15"^^xsd:date
"male"
rdf:type
foaf:name
vcard:BDAY
foaf:gender
?t
?p
?n
?b
?g
Result 1: {?t=univ:Student, ?p=univ:student123, ?n="John Green", ?g="male", ?b="1999-06-15"^^xsd:date}
Result 2: {?t=univ:Student, ?p=univ:student456, ?n="Susan Blue", ?g="female", ?b="2000-02-10"^^xsd:date}

?t
?p
?n
?b
?g
SPARQL Graph Pattern
Basic unit of SPARQL queries
rdf:type
foaf:name
vcard:BDAY
foaf:gender
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX foaf: <http://xmlns.com/foaf/0.1/>
PREFIX vcard: <http://www.w3.org/2001/vcard-rdf/3.0#>
SELECT ?t ?n ?b ?g
WHERE
{ ?p rdf:type ?t .
?p foaf:name ?n .
?p vcard:BDAY ?b .
?p foaf:gender ?g }
How do we express this with SPARQL?
Basic Graph
Pattern (BGP)

SPARQL FILTER: Restricting Solutions
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX foaf: <http://xmlns.com/foaf/0.1/>
PREFIX vcard: <http://www.w3.org/2001/vcard-rdf/3.0#>
PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>
SELECT ?n ?b ?g
WHERE
{ ?p foaf:name ?n .
?p vcard:BDAY ?b .
?p foaf:gender ?g
FILTER ( ?b < "2000-01-01"^^xsd:date )}
Find all people born before 2000

SPARQL Subqueries & Aggregates
SELECT DISTINCT ?n ?o ?p ?st ?cnt
WHERE
{ ?s foaf:name ?n .
?s vcard:BDAY ?b .
?s foaf:gender ?g
{ SELECT ?s (COUNT(?c) AS ?cnt)
WHERE
{ ?s univ:takes ?c }
GROUP BY ?s
HAVING (COUNT(?c) > 20)
}
}
ORDER BY DESC(?cnt) ASC(?n)
Find information about Students that have taken more than 20 classes

SPARQL Property Path: Matching Arbitrary Length Paths
SELECT ?friend
WHERE
{ ?student foaf:name "Bill Brown" .
?student foaf:friendOf+ ?friend
}
Find all the friends of Bill Brown
Regular expression-style syntax
Syntax Matches
:predicate+ Path composed of one or more repetitions of :predicate
:predicate* Path composed of zero or more repetitions of :predicate

SPARQL 1.1 Update
• SPARQL 1.1 Update is intended to be a standard language for specifying and
executing updates to RDF graphs in a Graph Store.
Graph Update operations
INSERT DATA Adds some triples, given inline in the
request, into a graph.
DELETE DATA Removes some triples, given inline in the
request, if the respective graph contains those.
DELETE/INSERT Perform pattern-based actions for graph
updates
LOAD Reads the contents of a document representing a
graph into a graph in the Graph Store.
CLEAR Removes all the triples in (one or more) graphs.
Graph Management operations
CREATE Creates a new graph in stores that support empty
graphs.
DROP Removes a graph and all of its contents.
COPY Modifies a graph to contain a copy of another.
MOVE Moves all of the data from one graph into another.
ADD Reproduces all data from one graph into another.

SPARQL and SPARQL in SQL Architecture
Support for Apache Jena
Standard SPARQL Endpoint
Enhanced with query management control
SEM_MATCH
SEM_APIS.UPDATE_MODELSQL
Java
HTTP
SPARQL-to-SQL Core Logic
Apache Jena

SPARQL in SQL: SEM_MATCH
SELECT n1, n2
FROM
TABLE(
SEM_MATCH(
'PREFIX foaf: <http://...>
SELECT ?n1 ?n2
FROM <http://g1>
WHERE {?p foaf:name ?n1
OPTIONAL {?p foaf:knows ?f .
?f foaf:name ?n2 }
FILTER (REGEX(?n1, "^A")) }
ORDER BY ?n1 ?n2',
SEM_MODELS('M1'),…));
n1 n2
Alex Jerry
Alex Tom
Alice Bill
Alice Jill
Alice John

SPARQL in SQL: SEM_MATCH
• Extends SQL with full SPARQL 1.1 query constructs
• Benefits:
– Integrates graph data with existing enterprise data
– JOINs with other relational (and object-relational) data
– Allows SQL constructs/functions
– DDL Statements: create tables/views
– Allows use of enterprise SQL development tools

SPARQL Update: SEM_APIS.UPDATE_MODEL
begin
sem_apis.update_model(
apply_model=>'M1',
update_stmt=>'INSERT {?s :mbox ?n}
WHERE {?s :email ?n}'
);
end;
/

Tools Make Things Easier
SQL Developer RDF Plugin Jena Fuseki Web Application

Agenda
64
1
2
3
4

Oracle RDF KG as an Integration / Analytics Platform
My SQL
SQL Server
DB2
Oracle Big
Data
Connectors
External
Table /
DB Link
Oracle
Database
Gateways
Oracle Advanced
AnalyticsOBIEE
Relational
Table/View
Property
Graph View
Property
Graph View
of RDF
RDF View
PGX
CREATE VIEW
SEM_MATCH

Creating Property Graph Views on an RDF Graph
public static void createPropertyGraphViewOnRDF(Connection conn,
String pgGraphName,
String rdfModelName,
boolean virtualModel,
PGUtils.RDFPredicate[] predListForVertexAttrs,
PGUtils.RDFPredicate[] predListForEdges)
This method creates the property graph views (vertex and edge views) for the given RDF model.
Parameters:
conn - a connection instance to Oracle database
pgGraphName - the name of the property graph to be created
rdfModelName - the name of the RDF model
virtualModel - a flag represents if the RDF model is virtual model or not
predListForVertexAttrs - an array of RDFPredicate objects specifying how to create vertex view using these
predicates
predListForEdges - an array of RDFPredicate specifying how to create edge view using these predicates

Creating Property Graph Views on an RDF Graph
// create PG view on RDF model
PGUtils.createPropertyGraphViewOnRDF(conn, "articles", "articles", false,
predsForVertexAttrs, predsForEdges);
// get the Property Graph instance
oracle = new Oracle(jdbcUrl, user, password);
pggraph = OraclePropertyGraph.getInstance(oracle, "articles");
// use like a normal Property Graph instance
System.err.println("------ Vertices from property graph view ------");
pggraph.getVertices();
System.err.println("------ Edges from property graph view ------");
pggraph.getEdges();

Questions?

Build Knowledge Graphs with Oracle RDF to Extract More Value from Your Data

Build Knowledge Graphs with Oracle RDF to Extract More Value from Your Data

Recommended

Recommended

More Related Content

What's hot

What's hot (20)

Similar to Build Knowledge Graphs with Oracle RDF to Extract More Value from Your Data

Similar to Build Knowledge Graphs with Oracle RDF to Extract More Value from Your Data (20)

More from Jean Ihm

More from Jean Ihm (10)

Recently uploaded

Recently uploaded (20)

Build Knowledge Graphs with Oracle RDF to Extract More Value from Your Data