Sematic Web Road map

1
Sematic Web Road map

• Fredrik Huitfeldt Madsen
• Mads Tjørnelund Toustrup
• Jens Erik Meldgaard Thomsen

2
Introduction

• The article is a road map for a concept called
  the semantic web.
• The article is only a draft, and is extremely
  difficult to read. If there are misunderstandings
  they are probably because of that.
• The semantic web is Tim Berners-Lees future
  version of the Internet.

3
The Road Ahead

• The problems with the current state of the
  Internet are that the machines cannot understand
  whats in it. When the machines cannot understand
  the data, theyre of little use helping people.
• If we want to improve this, we can either make
  the machines capable of understanding the current
  Internet, or we can change the Internet so that
  its more understandable to machines.

4
The Road Ahead - continued

• So in Tim Berners-Lees future version of the
  Internet, the data residing on the Internet is
  put in a form so that it is understandable to
  machines.
• The advantages of this are obvious the
  information contained in the Internet would be
  easily accessible, and noise could be efficiently
  filtered away.

5
The four steps

• For obvious reasons, this cannot be achieved over
  night, so Tim Berners-Lee have suggested dividing
  the path up into four steps, which can be
  implemented successively.
• The basic assertion model
• Enabling us to describe objects, and their
  relations.
• The schema layer
• Enabling us to describe classes.
• The conversion language
• Enabling us to convert objects between different
  classes.
• The logical layer
• Enabling us to make logical expressions across
  objects from different classes.

6
The basic assertion model

• When making a formulation of all data on the web,
  for obvious reasons it has to be extremely
  general. Therefore the basic assertion model
  should be based on the RDF, the Resource
  Description Framework.
• In the RDF, the world is described, using
  resources, properties, and expressions. Using
  object oriented design terminology this
  corresponds to objects, properties, and
  relations. Using the terminology of the article,
  properties are assertions about resources, and
  expressions are quotations.

7
The basic assertion model - cont.

• The basic assertion model gives us the
  opportunity to make assertions about the data
  described in the RDF.
• A is the author of X.

8
The basic assertion model - cont.

• It also enables us to make assertions about
  assertions, also called quotations. An example of
  a quotation is
• The author of X is American.
• At this very basic level, were able to describe
  objects, and their relations.

9
The Schema Layer

• The schema layer corresponds to what we know as
  the class diagram. It is a description of how the
  different objects of the same class are
  described. As with class diagrams, you can
  express that an attribute should be unique.

10
Conversion Language

• A conversion language is needed in order to
  convert objects described in one schema, to
  objects described in another.
• Ex
• Table A is described by the attributes A1 and A2
• Table B is described by the attributes B1 and B2
• If we know that there are objects in table A,
  which are also in table B, and we want to be able
  to use that knowledge, we need to be able to map
  these objects to each other. If we can register
  this knowledge, everybody will be able to use the
  total knowledge of the two tables.

11
Conversion Language - continued

12
The logical layer

• We need ways of writing logic into documents to
  allow such things as
• Rules of deduction of one type of documents from
  a document of another type
• The checking of a document against a set of rules
  of self-consistency
• The resolution of a query by conversion from
  terms unknown into terms known

13
The logical layer - continued

• A simple example of applications of this layer
• Two databases constructed independently
• Put on the web and linked together by semantic
  links
• This allows queries on one database to be
  converted into queries on another

14
The logical layer - continued

• When defining a new language for a specific
  application, two things are required
• Settle with the limited power of the reasoning
  engine of the receiver and define a subset of
  full RDF
• Define some abbreviate functions to transmit
  expressions

15
Metro map of the semantic web
16
Proof validation

• The RDF model does not say anything about the
  reasoning engine
• There is no perfect algorithm for answering
  questions or finding proofs
• The proof validation is done by some fairly
  constrained rules, and all that the other part
  has to do is validate a general proof

17
Proof validation - continued

• For example when someone is granted access to a
  web server, a document could explain the server
  why they should have access
• The proof in the semantic web is a chain of
  assertions and reasoning rules with pointers to
  all the supporting material
• This is like the HTTP GET, which contain a
  proof that the client has right to the response

18
Evolution Rules Language

• Two fundamental functions required of the RDF
  engines
• For a version n implementation to be able to read
  enough RDF schema to be able to deduce how to
  read a version n1 document.
• For a type A application developed quite
  independently of a type B application which has
  the same or similar function to be able to read
  and process enough schema information to be able
  to process data from type B application.

19
Example

• Library of Congress uses the word author of a
  book.
• British Library uses the word creator of a
  book.
• A small bit of RDF would be able to say
• If x it the author (LoC) of y
• Then x is also the creator (BL) of y.

20
Where would a processor find these rules?

• In the case of different versions
• The versions will contain the rules or have a
  pointer to them.
• In the case of different applications
• Each application could have the rules
• Each application could have a pointer to a third
  party index which can be search for connection.

21
Query Languages

• A query can be thought if as an assertion about
  the result to be returned.
• Example
• Query Author of a resource.
• Assertion X is the author of p1, for some X.
• Query List of all authors.
• Assertion Any author is in the set and everyone
  in the set is an author.

22
Digital Signature

• The common logical model requires extension to
  include the keys with which assertions have been
  signed.
• For proof validation
• For inference
• Check the logic
• For assertions that a document has been signed
• Check the signature

23
Indexes of terms

• The search engine technology currently applied to
  HTML pages will presumably translate directly
  into indexes not of words, but of RDF objects.
• This will result in much more efficient searching
  of the web.
• Search in one giant database, rather than one
  giant book.

24
Engines of the Future

• Search engines which index HTML pages
• Find many answers and cover huge parts of the
  web.
• Return many inappropriate answers
• No guarantee of correctness
• Logical engines
• Restrict their output (provably correct answer)
• Inability to rummage through the mass of
  intertwined data to construct valid answers.

25
Combine the two engines

• Take the best of each engine.
• Use logic to guarantee correctness on large
  indexes.
• Concluding remark of the article
• Though there will still not be a machine which
  can guarantee to answer arbitrary question, the
  power to answer real question which are the stuff
  of our daily lives and especially of commerce may
  be quite remarkable.