Co-funded by the European Union Semantic CMS Community Semantic Data Access Copyright IKS Consortium 1 SRDC Ltd. August, 2011

Page: Outline  Semantic Data  Semantic Web  RDF  Semantic Data Storage  Triple Stores  Semantic Data Access  SPARQL  RQL  API Calls Copyright IKS Consortium 2

Page: Semantic Data  Stands for machine understandable information  Allows computers to figure out the data without user interference  Allows computers act intelligently without programming for each task

Page: Semantic Data  Provides infrastructure to get practical results  Applications find out subsequent information based on the previous relations. (e.g Eiffel Tower -> Paris -> France)  Allows reasoning capabilities  Providing extraction of related information which is not directly linked

Page: Semantic Web  A classical generic description:  “Web of data”  Extends the World Wide Web  By encouraging,  Common language for representing data  Transformable to/from disparate sources such as relational databases, XML, etc (RDF)  Common reusable data model to represent data from different domains in common terms (RDFS, OWL, etc)  Rules to enable applications reason over the information (SWRL)

Page: Semantic Web Stack

Page: Semantic Web  So many organizations publishing their data in different domains  Media  Geographic  Government  …  Whole set contains approximately 30 billion triples  One of the largest collections is DBPEDIA  Semantified version of Wikipedia  Example:  Obtain cities of China that have population over 20 million  Needs efficient storage and query for semantic data Copyright IKS Consortium 7

Page: Representation of Semantic Data  RDF  The common data format  An abstract model with several serialization formats  Consists of statement referred as triples having the form (subject, predicate, object) where,  Subject: any resource identifier  Predicate: a resource identifier of any property  Object: either a resource identifier or a literal value

Page:  Two types of resource identifiers:  URIRef  BNode.  Property  Used when talking about the particular aspect of a resource.  Must be represented by URIRefs and may not be given BNode identifiers.  Literal  Used when the object of the statement has no resource identifier.  Represents statement itself RDF

Page: RDF Serialization Formats  RDF/XML  N3  N-Triples  TRiG  TRiX  Turtle  JSON  JSON-LD  RDFa

Page: RDF Serialization Formats  RDF/XML is one of the most used serialization format making use of  Relative and absolute URIs  Namespaces  XSD Datatypes  The terms declared in XML Information Set

Page: RDF Serialization Formats  Notation 3 (N3)  More human readable RDF serialization  Aims to integrate logic and data in the same language by allowing smooth integration of rules with RDF

Page: RDF Examples  RDF/XML <rdf:RDF xml:base=“ xmlns:rdf=" xmlns:dbpprop=“ xmlns:dbpont=“

Page: RDF Examples Cont’d  N3 notation of previous ex: ex:Japan rdf:Type dbpont:Country; dbpprop:areaKm ; dbpprop:city ex:Tokyo. ex:Tokyo rdf:Type dbpont:City

Page: Storing Semantic Data  Need for specialized designs for triple collections  Two modalities:  Relational databases  Triple stores  Mostly used for storage  Lots of implementations  They can also be RDB based.

Page: Triple Store  A purpose-built database for the storage and retrieval of RDF data.  Optimized place to add, remove and query for triples. Each triple in the TripleStore complies with the form (subject, predicate, object)

Page: Considering XML Databases  XML databases are existing storage systems for semi- structured data  Idea: Transform RDF to XML and store it in XML databases  Yet, XML data model is not exactly same with semantic data  XML data model is a tree-like structure  RDF data is represented through a graph without an hierarchy Copyright IKS Consortium 17

Page: Considering XML Databases  XML Databases are not suitable for storage and querying RDF  Only simple manipulations can be handled through XML query languages  RDF Schema processing and inference is not possible  Standard RDF/XML mapping is unsuitable Copyright IKS Consortium 18

Page: Monolithic approach for DB Based Triple Stores  Generic representation for all RDF schemas  Only two tables are used  Resources table  Triples table Copyright IKS Consortium 19

Page: Monolithic approach for DB Based Triple Stores Copyright IKS Consortium 20 predidsubidobjidobjvalue Sunscal e iduri 1http:// 2http:// 3http:// 4http:// 5http:// 6http:// 7http:// ns#Property 8http:// 9rl

Page: Triples Stores  Can be categorized into 3 category:  In memory triple stores  Used for certain operations like benchmarking, caching, etc  Native triple stores  Provides their own implementations (Virtuoso, Mulgara, AllegroGraph, …)  Non memory non native triple stores  Are built on third party databases (Jena SDB, Kaon, …)

Page: Functionalities provided by Triple Stores  RDBMS-support  General RDF model access  Query language support in the store such as RQL, SPARQL  Some stores provide:  Provenance - tracking of who-said-what  APIs for accessing triple store over network  Very few stores provide:  Full text search  Inference and rule languages Copyright IKS Consortium 22

Page: Example Triple Store implementations  RDF Suite  Sofia Alexaki, Vassilis Christophides, Gregory Karvounarakis, Dimitris Plexousakis, Karsten Tolle. The ICS-FORTH RDFSuite: Managing Voluminous RDF Description Bases, SemWeb, 2001  Based on an ORDBMS model  Sesame   Relational databases (mysql, postgres, oracle)  Jena   Relational databases (mysql, postgres, oracle)  Virtuoso   Native RDF Quad Storage (Physical Quads) Copyright IKS Consortium 23

Page: RDFSuite (ICS-Forth)* * IST C-Web, IST Mesmuses

Page: How triples are stored and accessed in RDF Suite  Separate tables are created to store resources  Properties, subClasses, subProperties and instances  Indices on attributes like URI, source and target  Querying is possible through RQL Copyright IKS Consortium 25

Page: How triples are stored and accessed in RDF Suite Copyright IKS Consortium 26 [ Figure from *] *Sofia Alexaki, Vassilis Christophides, Gregory Karvounarakis, Dimitris Plexousakis, Karsten Tolle. The ICS-FORTH RDFSuite: Managing Voluminous RDF Description Bases, SemWeb, 2001

Page: Sesame Architecture  DBMS-independent API for accessing triple repositories  SAIL API  A set of Java interfaces between other modules and repository  Abstract from the actual storage mechanism  Query Module  RQL support  Different ways to communicate with clients  Through Protocol handlers Copyright IKS Consortium 27 *Jeen Broekstra and Arjohn Kampman and Frank van Harmelen, Sesame: A Generic Architecture for Storing and Querying RDF and RDF Schema, Proceedings of the First International Semantic Web Conference, 2002

Page: SAIL API over PostgreSQL  PostgreSQL  Object-relational DBMS  Support sub-table relations between its tables for providing RDF Schema class and property subsumption  Individuals are represented under separate tables created for resources  Difficult to add table *Jeen Broekstra and Arjohn Kampman and Frank van Harmelen, Sesame: A Generic Architecture for Storing and Querying RDF and RDF Schema, Proceedings of the First International Semantic Web Conference, 2002

Page: SAIL API over MySQL  MySQL  The database schema does not change when the RDFS changes  Has advantage where RDFS is unstable *Jeen Broekstra and Arjohn Kampman and Frank van Harmelen, Sesame: A Generic Architecture for Storing and Querying RDF and RDF Schema, Proceedings of the First International Semantic Web Conference, 2002

Page: Jena2 Architecture Copyright IKS Consortium 30

Page: Jena2 Architecture Copyright IKS Consortium 31 *Kevin Wilkinson, Craig Sayers, Harumi A. Kuno, Dave Reynolds: Efficient RDF Storage and Retrieval in Jena2, Proceedings of SWDB'03, The first International Workshop onKevin WilkinsonCraig SayersHarumi A. KunoDave Reynolds Semantic Web and Databases

Page: Jena2  Jena2  Denormalized schema  Avoids unnecessary joins by merging URIs, literals in statements table  Multiple statement tables  Better locality and caching  Property Tables Copyright IKS Consortium 32

Page: Normalized vs Denormalized Tables Copyright IKS Consortium 33

Page: Property Tables Copyright IKS Consortium 34 SubjectPropertyObject person1nameAlice person1age32 person1twinOfperson2 person1faxPhonex1234 person1adminPhx5678 person2nameBob person2age35 person2adopteeOfperson6 person2friendOfperson8 person2gendermale SubjectPropertyObject person1twinOfperson2 person1faxPhonex1234 person1adminPhx5678 person2adopteeOfperson6 person2friendOfperson8 IDnameagegender p1Alice32- p2Bob35male Triple Store Person Property Table Triple Store Only *Kevin Wilkinson, Craig Sayers, Harumi A. Kuno, Dave Reynolds: Efficient RDF Storage and Retrieval in Jena2, Proceedings of SWDB'03, The first International Workshop onKevin WilkinsonCraig SayersHarumi A. KunoDave Reynolds Semantic Web and Databases

Page: Jena Persistence Options  SDB  Scalable storage and query for RDF  Specifically designed for SPARQL support  Supports: MySQL, PostgreSQL, Oracle 11g, Microsoft SQL server and IBM DB2  Scales to graphs of 100 million triples Copyright IKS Consortium 35

Page: Jena Persistence Options  TDB  Provides for large scale storage and query of RDF datasets using a pure Java engine  Supports SPARQL  A non-transactional, faster database solution for use by a single system  It scales well beyond SDB and is simpler to setup Copyright IKS Consortium 36

Page: Virtuoso  General purpose RDBMS with extensive RDF adaptations  RDF data is stored as RDF quads, i.e. it supports RDF with named graphs  i.e. graph, subject, predicate, object tuples  The columns are G for graph, P for predicate, S for subject and O for object Copyright IKS Consortium 37

Page: Querying Semantic Data  Semantic data can be queried from triple stores by  Various query languages  SPARQL  Different endpoints provided  RQL  RDQL  SeRQL  …  API Calls  Through proprietary APIs of different projects  Linked Data

Page: SPARQL  Is an RDF query language  Standardized by W3C Concortium  Similar concept of SQL for databases  Syntactically resembles to SQL  RDF Graphs instead of databases

Page: SPARQL  Provides 4 types of query:  SELECT  Used for querying RDF graph by selecting certain fields from the query pattern  CONSTRUCT  Used for constructing a single RDF graph specified by a graph template  ASK  Used for querying existence of a resource  DESCRIBE  Used for construction an RDF graph, but the structure of graph determined by SPARQL query processor unlike Constructor type

Page: SPARQL  SPARQL SELECT queries can be constructed through the following parts:  Prefix declaration  Field declarations  Dataset selection  Query pattern  Query modifiers

Page: Prefix Declaration  Prefix declarations are specified to be able to use short URIs instead of the full ones PREFIX rdf:. PREFIX foaf:.

Page: Field declarations  Desired fields from the query pattern are specified after SELECT keyword with PREFIX rdf:. PREFIX foaf:. SELECT ?person ?name

Page: Dataset selection  RDF graphs to be queried are stated after together with a FROM clause PREFIX rdf:. PREFIX foaf:. SELECT ?person ?name FROM

Page: Query Pattern  A set of triples that determines the target resources to be selected PREFIX rdf:. PREFIX foaf:. SELECT ?person ?name FROM WHERE { ?person foaf:name ?name. }

Page: Query Modifiers  There are a plenty of query modifiers in SPARQL providing post processing of query results  Order  Limit  Distinct  Offset  Reduced  Projection

Page: Query Modifiers  They are used to  Eliminate duplicate results  Ordering the results  Limiting the number of returned results, etc PREFIX rdf:. PREFIX foaf:. SELECT ?person ?name FROM WHERE { ?person foaf:name ?name. } ORDER BY ?name LIMIT 5 OFFSET 10

Page: FILTER and OPTINAL CLAUSES  FILTER provides calling a subset of functions provided by XQuery specification in query pattern of SPARQL PREFIX rdf:. PREFIX foaf:. PREFIX fn:. SELECT ?person ?name FROM WHERE { ?person foaf:name ?name. FILTER (fn:string-length(?name) > 10) }

Page: FILTER and OPTINAL CLAUSES  OPTIONAL clause enables to specify query patterns of which match are not obligatory in query execution. PREFIX rdf:. PREFIX foaf:. PREFIX fn:. PREFIX info:. SELECT ?person ?name FROM WHERE { ?person foaf:name ?name. FILTER (fn:string-length(?name) > 10) OPTIONAL { ?person foaf:homepage ?page. } }

Page: ASK Query Example PREFIX foaf:. ASK WHERE { ?person foaf:name “Tim Berners-Lee”. }  Check existences of a resource having “Tim Berners- Lee” as foaf:name.

Page: CONSTRUCT Query Example  Below example construct an new RDF graph by changing type values of skos:Concept resources. In other words, that means transformation of skos vocabulary to a new custom vocabulary PREFIX rdf:. PREFIX skos:. PREFIX myvocab:. CONSTRUCT { ?person rdf:Type myvocab:MyType. } WHERE { ?person rdf:Type skos:Concept. }

Page: SPARQL Endpoints  Provides functionality to query the knowledge base via the SPARQL language  Accepts queries and returns results through HTTP protocol  Query results can be in different formats such as  RDF  XML  HTML  JSON  CSV

Page: Semantic Data Access With API Calls  Open source projects provides APIs to manipulate RDF data  Jena  Apache Clerezza  Sesame  JRDF

Page: Jena  Jena provides a rich API to manipulate the RDF stored in the underlying triple store.  Model to represent graphs  CRUD methods for triples  Querying methods for existing resources  See the next slide for the code snippet…

Page: Jena Code Snippet String personURI = " String givenName = "John"; String familyName = "Smith"; String fullName = givenName + " " + familyName; // create an empty Model which represents an RDF graph Model model = ModelFactory.createDefaultModel(); // create the resource which will produce the triples in the next slide Resource johnSmith = model.createResource(personURI).addProperty(VCARD.FN, fullName).addProperty(VCARD.N, model.createResource().addProperty(VCARD.Given, givenName).addProperty(VCARD.Family, familyName));

Page: Jena  Created triples with the code snippet in previous slide: (, VCARD.FN, “John Smith”) (, VCARD.FN, _) (_, VCARD.Given, “John”) (_, VCARD.Family, “Smith”) Note that _ symbol represents a blank node

Page: Apache Clerezza  Provides an API regardless from the different triples stores it supports  Its API provides a model to represent RDF graphs and manipulate those graphs  Also provides an SPARQL endpoint to query the stored knowledge

Page: Apache Clerezza Code Snippet String base = “ MGraph g = new SimpleMGraph(); g.add( new TripleImpl( new UriRef(base + “JohnSmith”), new UriRef(rdf:Type) new UriRef(foaf:Person))); g.add( new TripleImpl( new UriRef(base + “JohnSmith”), new UriRef(VCARD:FN) LiteralFactory.getInstance().createTypedLiteral(“John”)));  Simple code snippet adding two triples to the graph:

Page: Linked Data  Interrelated datasets on the Web so that computers can explore them  Has a standard format to be accessed and managed  Provides integration and reasoning on a huge amount of data on the Web

Page: Linked Data  Four famous principles of linked data represented by Tim Berners-Lee  Use URIs as names of things  Use HTTP URIs to provide dereferencable data to people  When an URI is dereferenced provide useful information in standard format (RDF, SPARQL)  Provide links to other URIs to make possible discovery of related data

Page: Linked Data

Page: Linking Open Data Project  Is an W3C SWEO Project  Aims to make data freely to everyone  Aims to publish open data sets as RDF and set semantic relationships between them  Serves information in a machine readable format  Enriches content  Reduces duplication  Linked datasets increasing rapidly  A large number of datasets are linked already

Page: Linked Datasets As of October 2008

Page: Linked Datasets As of September 2010

Page: 2011

Page: Access Data In The Cloud  Follow the RDF links representing the “things”  SPARQL Endpoints  Ready to use software to discover linked data (See the next slide)

Page: Linked Data Applications  Lots of application on top of the linked data  Tabulator  Marbles  Openlink RDF Browser  …  Just google  RDF Crawlers  RDF Browsers  Also see the following link containing a number of linked data applications:  LinkingOpenData/Applications LinkingOpenData/Applications

Page: Available SPARQL Endpoints    To see possible SPARQL endpoints providing a certain URI see   See also a list of alive SPARQL endpoints 

Page: References     web?src=related_normal&rel= web?src=related_normal&rel=      Sofia Alexaki, Vassilis Christophides, Gregory Karvounarakis, Dimitris Plexousakis, Karsten Tolle. The ICS-FORTH RDFSuite: Managing Voluminous RDF Description Bases, SemWeb, 2001  Jeen Broekstra and Arjohn Kampman and Frank van Harmelen, Sesame: A Generic Architecture for Storing and Querying RDF and RDF Schema, Proceedings of the First International, Semantic Web Conference, 2002  Kevin Wilkinson, Craig Sayers, Harumi A. Kuno, Dave Reynolds: Efficient RDF Storage and Retrieval in Jena2, Proceedings of SWDB'03, The first International Workshop on Semantic Web and Databases  