Agents and the Semantic Web By James Hendler

1
Agents and the Semantic WebBy James Hendler

• Presented by
• Yufei Xu
• Huaicheng Su

2
Outline

• Related concepts covered in this topic
• Why the employment of multi-agent system in
  Semantic web is possible and practical
• page creation with ontological information
• definition of web services in machine readable
  format
• communication between multi-agents
• Summarization and Conclusion

3
Related concepts

• Multi-agent system
• Definition A computational environment in
  which individual software agents interact with
  each other, sometime in a cooperative manner, and
  sometimes autonomously pursuing their individual
  goals, accessing resources and services of the
  environment, and occasionally producing results
  for the entities that initiated those software
  agents.
• Hence, in such a system, computer programs
  (agents) can collect and process environments
  resources and exchange information freely.

4
Related Concepts (Cont.)

• Semantic Web
• W3C Definition The semantic web is a
  vision the idea of having data on the web
  defined and linked in a way that it can be used
  by machines not just for display purposes, but
  for automation, integration and reuse of data
  across various applications.

5
Related Concepts (Cont.)

• A problem arises
• Does semantic web provide a suitable
  environment for applying multi-agent system?
  -----Answer is Yes.

Ontology
6
Related Concepts (Cont.)

• Ontology
• Definition A set of knowledge terms,
  including the vocabulary, the semantic
  interconnection and some simple rules of
  inference and logic for a particular topic.
• For example the ontology of cooking and
  cookbooks
• could include
• Vocabulary a set of ingredients
• Semantic interconnection how to stir and combine
  them
• Simple rule the difference between simmering and
  deep-frying the expectation of the dishes being
  eaten.

7
Related Concepts (Cont.)

• Incremental Ontology Creation
• Instead of large, complex, consistent ontology
  shared by many users, the semantic web has large
  number of smaller ontological components.
• Hence, we can construct ontology by links to
  others for reusing and changing terms.
• case study Mary is building an ontology for
  her online pet shop.
• She may visit the ontology repository and
  select ones of her interest. Based on those, she
  make some restriction and modification that
  satisfies her business characteristics. Finally,
  she may put her ontology back to the repository
  for sharing.

8
Semantic Web Creation

• The crucial aspect of creating the semantic web
  is to enable users who are not ontological
  experts to create machine readable web contents.
• Only a small number of tool creators and web
  ontology designers need to know the details.
• Ideally, for most users, they shouldnt even know
  the ontology exists when creating the web pages.
  In other words, the semantic markup cost should
  be free.

9
Semantic Web Creation (Cont.)

• To fully understand Markup free concept, an
  illustrative example may be helpful.
• Consider a specific page creation tool driven
  by ontologies
• 1. Represents hierarchical class in menus
• 2. Properties of class could have many forms

10
Semantic Web Creation (Cont.)

• Then, a user who wants to add information about
  daughter can select the daughter class from the
  menu.
• The system will ask user to fill out all
  properties specified in the daughter class maybe
  through a form or browser to include links to
  home page or images.
• After collect all these instance information,
  it will use a appropriate web page design tool to
  lay out this information and add the markup to
  the ontology.
• As this tool can be driven by any ontologies, so
  we can create various pages such as homepage by
  using ontologies relating to Person and Hobby,
  professional pages by using ontologies relating
  to professional occupation or industries and so
  forth.

11
Semantic Web Creation (Cont.)

• From the example, we can know that through an
  interactive way, the tool would help a user to
  create a page and would provide free markup.
• On the other hand, the semantic web ontology
  could be build up incrementally, with links to
  many ontological components. Therefore, it
  enables to create semantic web of pages linking
  to many different ontologies.

12
Associate Web Service with Semantics

• Semantic web is an extension of the current web
• As a crucial component of the current web, web
  service in semantic web should also be defined in
  a machine readable format.
• So, we should associate it with semantics by
  using ontological language

13
Associate Web Service with Semantics

• The following graph may give us a vision of how
  to create semantic web services

14
Associate Web Service with Semantics

• A specific service such as weather service may be
  a subclass from a general service class.
• We also define some properties on the web
  service
• Advertisement property for service locating
• Description property for invocation
  description
• Service logic property for consequence
  indication
• Service advertisement
• By using ontological language, we can define
  a machine readable advertisement. Hence, lead to
  a better searching results.
• e.g. the machine/agent can follow the class
  hierarchy or other property restrictions to find
  the matching which is an improvement compared
  with the key-word based search.

15
Associate Web Service with Semantics

• Service invocation description
• This property describes the scenario
  regarding to invoke a web service.
• e.g. Consider invoking a buying book
  service, you may lead to filling out a sequence
  of web forms to provide all the information
  necessary to make an order.
• This kind of scenario is commonly implemented
  by writing a procedure code which calls various
  functions.
• In other words, the invocation procedure is
  represented procedurally on the web.

16
Associate Web Service with Semantics

• Service invocation description (Cont.)
• This results a problem an agent coming to a
  service cant analyze the necessary information
  required to invoke it. Because it only has a
  local view of a specific step and cant get the
  global view without actually invoking all the
  steps.
• This problem can be solved using ontological
  language to define a service invocation
  description.
• e.g. A possible technique can be adopted
  here We can describe the invocation by using
  FSM( Finite State Machine) where each step is
  represented as a state and use links between them
  to represent the transition.

17
Associate Web Service with Semantics

• A base ontology for this FSM looks like the
  following

18
Associate Web Service with Semantics

• Service invocation description (Cont.)
• The service provider can extend the above
  base FSM ontology to define his own FSM for
  expressing invocation description of his specific
  service.
• Doing this, provides agents a global view of
  invoking the service.
• What more important is that it is machine
  readable, so agent can analyze the invocation
  description to find the necessary information
  required for invoking the service.

19
Associate Web Service with Semantics

• Service Logic
• The service logic property describes the
  consequence of using a service. This can be used
  to express information that goes beyond what is
  contained in the service description.
• e.g. Users often leave a site without
  completing some steps in the transaction. So,
  they can not always know whether they are
  involved in a credit card charge. Using service
  logics, such things can be made explicit.

20
Associate Web Service with Semantics

• Service Logic (Cont.)
• sometimes more interesting transactional
  logic can be used in a proof checking to confirm
  transaction between agents

21
Associate Web Service with Semantics

• Service Logic (Cont.)
• Such service logic can also serves for many
  other purpose
• e.g. for planning a set of services that
  together achieve a users
• goal.
• can be tied to appropriate service
  description to indicate what
• an agent can do and when it can do or
  cant do so.
• After covering all aspects of defining web
  service with semantics, an example may illustrate
  their overall usages
• Suppose a crew of fish men at sea
  unfortunately encounter an epic
• storm, the captain goes to an agent-enabled
  geographical server
• site, querying for Get me a satellite photo
  of this region .

22
Associate Web Service with Semantics (Cont.)
23
Multi-agent Communication in Semantic Web

• Having the pages, service descriptions and agent
  programs linked to many ontologies enables the
  communication between agents.

24
Multi-agent Communication in Semantic Web (cont.)

• In this graph
• Small box agents
• Large box ontologies contains terms used by
  agents
• Links provides mapping from one ontology to
  anther
• The communication between agents using the same
• ontology is straightforward.
• For agents using different ontologies, they
  communicate
• by mapping on the common terms they share.
• For simple mapping, e.g. inclusion, we can
  guarantee that every agents can communicate.

25
Multi-agent Communication in Semantic Web (cont.)

• However, for more efficient ontology mapping
  scheme, it is still be in research by those
  interesting in bring multi-agent system to
  semantic web.

26
Summarization and Conclusion

• In this presentation we covered
• Related concepts e.g. multi-agent system,
  semantic web and ontology.
• Provide you visions that ontology plays a key
  role in this topic.
• It enables to create machine readable web
  contents.
• It provide a way to define web service in an
  agent processible format.
• It facilitates the communication between
  agents.
• Since all these pieces are coming, we claim that
  bring multi-agent system to the semantic web is
  possible and practical.

27
Question ?