Fall/Spring Horus Effort: Strawman

1
HORUS
The Egyptian All-Seeing God of Light
DAML PI Meeting, Naushua, NH 17 Jul 2001
Don Conklin
A Joint IMO/DARPA Project
2
Horus Goals

• Provide simpler and more focussed access to
  unstructured (web pages and documents) and
  structured (relational) data
• By employing advanced text markup methods and
  schema to ontology mappings
• Tie information together across different
  Knowledge Domains and provide capability to
  follow threads of logic
• By employing ontologies to describe these
  Knowledge Domains
• Allow access to information organized by
  substantive content rather than access to
  documents about the content
• Shift away from a document centric to an object
  centric view of information
• Leverage the efforts of the DARPA Agent Markup
  Language (DAML) effort and W3C and other industry
  efforts (including XML)
• Use DAML tools and technologies as the basis
  Horus
• Feedback to DAML on what works and what doesnt
• Horus is not competitive but rather complementary
  to efforts using XML to mark up documents

3
Building a Semantic Web

• Ontologies
• Allow semantically organization and logical
  structure for knowledge domain(s)
• How data is initially viewed and selected for
  markup
• How data is queried upon and navigated through
• Horus follows the DAML paradigm of many small
  linked ontologies
• Follow the XML / RDF / DAML construct
• Building ontologies is not trivial
• Using mostly DAML tools for construction,
  validation, etc.
• Horus is building tools for importing ontologies
• Horus is building a set of five Toolkits
• Ontologies
• Markup of documents, manual and automated
• Structured data source access
• Portal
• Knowledgebase (for persistence, query)

4
Horus Vision

• DATA
• Structured
• Unstructured

Horus Enabled Portal
5
A Horus Example
Create and Link Horus knowledge objects

• Independently created knowledge objects
• Possibly by different production units

Car Bomb Event (discrete ID) Location
Macedonian Border Weapon 200lbs TNT Date
25Jun01 Perpetrator Organization

• Supported by their own web sites and databases
• Linked by continuously updated object/indices

Making the best use of the data we have
Ahmed Kahlid Nasr Citizen of Kosovo Member
of KLA Acts Involved in Car Bomb
KLA Location Kosovo Leader
SheikNasrallah Backing Iran
6
Implications of Horus Use

• Markup takes some effort by producers of
  information, and consumers (if allowed)
• Producers can benefit from markup as they are the
  first consumers
• There is a critical mass required for the markup
  to become helpful
• Business processes have to be established by
  management
• To define who can create KOs (authoritative
  producer)
• Who can modify these KOs (original and other
  producers)
• Who can create additional links between KOs
  (producers and consumers)
• Web site development can become using Horus
  Knowledge Objects to develop products
• Instead of producing more documents

7
Horus-Enabled Web Sites

• Organized around knowledge objects (KOs)
  representing the sites knowledge domain(s) based
  on the ontologies used
• Support the creation of appropriate KOs
• Allows the capture of pertinent data to populate
  the KO properties
• Allow expressing the relationships between KOs
• Has Horus tools integrated into its processes to
  allow
• Markup of documents, during or after their
  production
• Map database schema to ontologies for structured
  data access
• Has access to a Knowledgebase
• For persistent store of KOs
• For query execution
• Maintains links into data bases and documents to
  indicate support of KO property values
• Source document/database URIs are kept with the
  KO in the Knowledgebase
• Allow link back to the original document and/or
  drilldown to a database product
• Has tools to allow the easy construction of
  queries and navigation of KOs

8
Benefits of Horus-Enabled Web Sites

• Greatly improved access to knowledge about domain
  objects of interest
• Greatly improved organization of knowledge
• Easier browsing
• Higher quality searching
• Changes the focus of users from viewing
  documents to viewing knowledge
• Incorporates easily into production and web site
  development initiatives (low cost of entry)
• Technical brief follows