Diapositiva 1

1
OKKAM AN ENTITY NAME SYSTEM (ENS) FOR THE
SEMANTIC WEB FP7-215032 Large-scale Integrating
Project
Paolo Bouquet University of Trento, Italy Project
Coordinator
Semantic Interoperability A Practical
Approach Berlin, 25 March 2008
2
Factsheet
3
The OKKAM consortium

• University of Trento, IT
• L3S Hannover, DE
• SAP Research, DE
• Elsevier, NL
• Expert System, IT
• Europe Unlimited, BE
• MAC, IE
• EPFL, CH
• DERI Galway, IE
• University of Malaga, SP
• INMARK, SP
• ANSA, IT

5 universities
4 SMEs
3 corporations
4
Background a vision for the Semantic Web

• Knowledge representation is a field which
  currently seems to have the reputation of being
  initially interesting, but which did not seem to
  shake the world to the extent that some of its
  proponents hoped.
• It made sense but was of limited use on a small
  scale, but never made it to the large scale. This
  is exactly the state which the hypertext field
  was in before the Web.
• The Semantic Web is what we will get if we
  perform the same globalization process to
  Knowledge Representation that the Web initially
  did to Hypertext. We remove the centralized
  concepts of absolute truth, total knowledge, and
  total provability, and see what we can do with
  limited knowledge.
• Tim Berners-Lee, What the Semantic Web can
  represent, 1998

5
In practice
Semantic Web
Sonia
Knows
Paolo
Is_involved_in
Works-for
UniTN
Coordinates
UniMORE
Wisdom
Works-for
Web_page
Web_page
Web_page
www.unitn.it
www.trento.it
www.ryanair.com
href
href
href
www.l3s.org
href
href
www.paolobouquet.net
href
href
href
www.google.com
ockham.org
Hypertext Web
6
The Semantic Web as a global K space

• The Semantic Web way to information integration
• People will create and publish collections of RDF
  statements about any resources (documents,
  people, locations, events, topics, ).
• If information is contained in e.g. databases,
  they can be exported to RDF with some additional
  work.
• Any collection of RDF triples defines a graph,
  whose nodes are resources and arcs are relations
  between resources.
• The meaning of relations can be defined in
  vocabularies/ontologies
• Different RDF graphs can be merged by
  collapsing nodes and relations with identical
  identifier (URI).
• The expected outcome is a smooth information
  integration across distributed information
  sources (a global distributed space of knowledge)?

7
A Web-like scenario
London
8
London!?

• Did you mean
• London, KY
• London, Laurel, KY
• London, OH
• London, Madison, OH
• London, AR
• London, Pope, AR
• London, TX
• London, Kimble, TX
• London, MO
• London, MO
• London, London, MI
• London, London, Monroe, MI
• London, Uninc Conecuh County, AL
• London, Uninc Conecuh County, Conecuh, AL
• London, Uninc Shelby County, IN
• London, Uninc Shelby County, Shelby, IN
• London, Deerfield, WI
• London, Deerfield, Dane, WI

• Or
• London, Jack2612 Almes DrMontgomery, AL(334)
  272-7005
• London, Jack R2511 Winchester RdMontgomery, AL
  36106-3327(334) 272-7005
• London, Jack1222 Whitetail TrlVan Buren, AR
  72956-7368(479) 474-4136
• London, Jack7400 Vista Del Mar AveLa Jolla, CA
  92037-4954(858) 456-1850
• ...

9
Entity-centricIM
10
An RDF scenario
hotel.it
I-ESA2008.org
Marriot Hotel
built_in
located_in
I-ESA
Berlino
2000
held_in
attends
part_of
bouquet
Berlin
Germania
placed_in
rdftype
Germany
country
11
An example
Mariott Hotel
I-ESA
built_in
located_in
held_in
Berlin
2000
attends
part_of
Germany
bouquet
rdftype
country
Query find me an a hotel located where I-ESA is
held in 2008
12
Knowlegde integration in the global space

• To implement this scenario, however, we need to
  address and solve two serious integration
  problems
• Schema-level heterogeneity matching different
  vocabularies/ontologies so that equivalent (or
  related) classes and properties can be put in
  relation or collapsed
• Instance-level mismatch resolving identities
  between instances, so that all statements about
  the same entity can be found and used

13
State-of-the-art

• Schema-level integration is well-studied and
  supported (see ontology matching methods and
  tools)
• See presentations today!
• Instance-level integration is mostly neglected,
  perhaps because most people believe it is an easy
  (or at least easier) problem, mainly
  technological and not worth investigating for
  researchers.

14
And indeed

• the situation of entity-level integration is
  currently as follows
• The RDF metadata of some of the most important
  WWW and Semantic Web conference are poorly
  integrated at the instance (entity) level
• FOAF profiles rely on ad hoc methods for
  identifying people (and dont identify much else)
• The most common available ontology editors (e.g.
  Protégé) or metadata management systems (e.g. in
  digital libraries) generate local URIs for any
  newly created instance
• Some efforts exist to create reusable URIs (e.g.
  LSID, DOI), but are very vertical or commercial
• URI retrieval and reuse in general is not well
  supported

15
What we have today

• lthttp//www.w3.org/RDF/Validator/run/1198080996828
  megt lthttp//xmlns.com/foaf/0.1/namegt "Heiko
  Stoermer"
• lthttp//www.w3.org/RDF/Validator/run/1198080996828
  megt lthttp//xmlns.com/foaf/0.1/knowsgt
  _jA64256
• _jA64256 lthttp//xmlns.com/foaf/0.1/namegt "Paolo
  Bouquet"
• lthttp//www.owl-ontologies.com/Ontology1198072971.
  owlPerson_2gt lthttp//www.owl-ontologies.com/Ontol
  ogy1198072971.owlNamegt "Paolo
  Bouquet"lthttp//www.w3.org/2001/XMLSchemastring
  gt
• lthttp//www.owl-ontologies.com/Ontology1198072971.
  owlPerson_2gt lthttp//www.owl-ontologies.com/Onto
  logy1198072971.owlattendedgt lthttp//www.owl-onto
  logies.com/Ontology1198072971.owlConference_1gt
• lthttp//www.owl-ontologies.com/Ontology1198072971.
  owlConference_1gt lthttp//www.owl-ontologies.com/O
  ntology1198072971.owlNamegt "SWAP2007"lthttp//ww
  w.w3.org/2001/XMLSchemastringgt .

16
What we would like for tomorrow

• lthttp//www.okkam.org/entity/ok5f23a5cea6834c4dae7
  3b78cdc17aec1gt lthttp//xmlns.com/foaf/0.1/namegt
  "Heiko Stoermer"
• lthttp//www.okkam.org/entity/ok5f23a5cea6834c4dae7
  3b78cdc17aec1gt lthttp//xmlns.com/foaf/0.1/knowsgtlt
  http//www.okkam.org/entity/ok20070630118579125205
  6gt
• lthttp//www.okkam.org/entity/ok2007063011857912520
  56gt lthttp//xmlns.com/foaf/0.1/namegt "Paolo
  Bouquet"
• lthttp//www.okkam.org/entity/ok2007063011857912520
  56gt lthttp//www.owl-ontologies.com/Ontology119807
  2971.owlNamegt "Paolo Bouquet"lthttp//www.w3.or
  g/2001/XMLSchemastringgt
• lthttp//www.okkam.org/entity/ok2007063011857912520
  56gtlthttp//www.owl-ontologies.com/Ontology1198072
  971.owlattendedgt lthttp//www.okkam.org/entity/ok
  asdf12191185791252056gt
• lthttp//www.okkam.org/entity/okasdf121911857912520
  56gtlthttp//www.owl-ontologies.com/Ontology1198072
  971.owlNamegt"SWAP2007"lthttp//www.w3.org/2001/
  XMLSchemastringgt

17
The OKKAM approach from DNS to ENS

• By analogy with the Domain Name System (DNS) for
  the Web, we need something like an Entity Naming
  System (ENS) for the Semantic Web
• Instead of resolving domain names into IP
  addresses, the ENS will resolve local entity
  names (or descriptions) into their global unique
  identifiers (URI)
• It can be made available (through public APIs and
  protocols) as a service which can be invoked by
  any content creation and management application

18
ENS-empowered applications

• Applications which can interact with the ENS for
• Creating new content (in any format) in which
  entities are annotated with global IDs
• Offering smart services based on the availability
  of global IDs
• Examples text processors, HTML/XML editors,
  RDF/OWL editors, filters for exporting relational
  DBs, metadata management tools,

19
An example word processing
I-ESA2008 in Berlin
ENS
I-ESA2008 in ltidgtBerlinlt/idgt
20
What counts as an entity in OKKAM?

• OKKAM is about instances (individuals, objects),
  not about universal resources (e.g. classes or
  properties)?
• Why? Because schemas embody viewpoints, entities
  dont expect objections on this!
• For example
• Michael Schumacher, his web page, Ferrari,
  Maranello, Italy are all entities
• Pegasus, p, ?2, are entities
• Computer_science as a topic may be an entity
  (borderline )?
• Pizza Margherita in a food ontology ??
• La Divina Commedia, MS Word, VW Golf are
  tricky entities
• The class Person and the property
  Has_email_address are not entities
• Types of entities to start with people,
  locations, organizations, events.

21
The steps of OKKAM

• First, creating a scalable and sustainable
  infrastructure ENS for the Semantic Web, which
  will enable in practice a space of globally
  unique reusable identifiers (e.g. URIs)
• Second, bootstrapping the ENS by populating it
  with billions of entities and by making available
  tools for creating OKKAMized content
• Third, showcase the benefits of the approach by
  building high-impact applications and services
  which take advantage from this infrastructure to
  do something that was not possible before

22
Research challenges

• Infrastructure
• Designing and implementing a scalable,
  distributed and secure storage service
• Providing reliable and efficient methods for
  entity matching
• Providing methods for managing the lifecycle of
  entities
• Bootstrapping the Web of Entities
• Developing methods for automatically populating
  the repository of the ENS
• Supporting the OKKAMization of a critical mass of
  content and data
• Make available simple ENS-Empowered tools for end
  users
• Ensuring trust through secure and
  privacy-respectful operations

23
OKKAM applications

• In the project, there are three application
  scenarios
• Enterprise knowledge management systems Product
  Lifecycle Management (with and within SAP)
• Entity-intensive content production supporting
  the creation of news (ANSA) and scientific
  articles (Elsevier) in an OKKAM-empowered
  environment
• Entity-centric semantic search engine making
  search a different experience (DERI Ireland)

24
Expected advantages

• Improving data-level integration of information
• Automatic merging of knowledge from different
  sources on the Web
• Integration of corporate knowledge across
  formats, systems, applications
• Enabling distributed queries on multiple sources
• Web2.0
• Improving the quality of tagging of multimedia
  material
• Making social network portable across multiple
  platforms
• New tools for navigating/searching
• entity-centric semantic search
• entity-centric browser or explorers
• Vertical application domains
• business intelligence
• publishing news
• knowledge management
• ...

25
Conclusions

• OKKAM supports entity-centric semantic
  integration and interoperability
• First, find the entities you are interested it
• Second, collect all statements about them
• Third, clean results through schema/vocabulary
  mapping
• It provides a simple but powerful way of
  integrating data without starting from
  schemas/vocabularies
• It enables the vision of the Semantic Web as an
  open, decentralized, virtually integrated
  knowledge base sketched by TBL in the initial
  quote.

26
But why OKKAM?

• Ockham's Razor (14 century)
• entities should not be multiplied beyond
  necessity
• OKKAMs Razor (21 century)
• entity identifiers should not be multiplied
  beyond necessity

27
fp7.www.
.org