Transcript ppt
DARPA Agent Markup Language
Ashish Jain
University of Colorado at Boulder
What is DAML ?
Technology to enable software agents to
dynamically identify and understand
information sources.
Formatting web so that it can easily be
understood by intelligent agents
Express ontologies (formal specification of a
concept – vocabularies , inter-relationships etc)
Add reasoning cues
– Disjoint from , union of , transitive
DAML: Status
Currently being explored at University
level
– SHOE (Maryland) , OWL (Washington)
– Largely grows from past DARPA projects
But not transitioning
– W3C focused on short-term gains:
HTML/XML
DAML: Objectives
1.
Create an Agent Mark-Up Language (DAML)
built upon XML that allows users to provide
machine-readable semantic annotations for
specific communities of interest.
2.
Create tools that embed DAML markup on to
web pages and other information sources in a
manner that is transparent and beneficial to the
users.
3.
Use these tools to build up, instantiate, operate,
and test sets of agent-based programs that
markup and use DAML
DAML vs HTML vs XML vs RDF
HTML:
– Limited set of tags , not suitable for search
XML:
– Extensible tags
– Useful for data sharing
• but still not good for searching
RDF:
– We can only define global range on properties ( i.e. for
all classes apply constraints)
– No mechanism for providing necessary and sufficient
conditions
– No support transitivity
DAML: Example
1.
Find information about a researcher
named James Hendler
2.
Find a reference to a paper about SHOE
coauthored by James Handler
3.
Find a reference to the most recent
paper about SHOE coauthored by James
Handler
Query Processing (I)
DAML ontologies for publication,
researchers and topic have been built.
First query can be formalized as :
<Xmlns: SRIRes =
http://ai.src.com/daml/ontologies/Reseachers#>
<SRIRes: Reseacher>
<firstname>James</firstname>
<lastname>Hendler</lastname>
<title>Dr.</title> </SRIRes:Reseacher>
Query Processing (II)
After parsing the query , it is passed to
DAML-Q ( DAML query engine)
Looks for the namespace identifier, and
sequentially searches the content of the
web pages.
Could be complicated because of
presence of indirect ontological reference.
Inference in Queries (I)
Uses first order theorem prover such as
SNARK
Written in LISP
Inference in Queries (II)
Third query can be formalized as:
(find)
?paperq
such-that
(and
(pub-val ?paperq ?paper
(author ?paper ?person
(person-val (personq “James Hendler” ) ?person )
(about paper (paperq “SHOE”))
(= (pub-to-year ?paper) (year-fn ?natural)))
prefer
starts-after-starting-of
on
(year-fn ?natural)
Time-limit 10)
Conclusions (I)
Allows semantic interoperability at the level we
currently have syntactic interoperability in XML
– Revolutionizing web interoperability
Objects in the web can be marked (manually
or automatically) to include the following:
– Description of data they contain
– Description of function they provide
– Description of data they provide
This marks the environment for agents
Conclusion (II)
Lack of user tools to create it
Lack of agents that understand it
But step in right direction