The ABC Metadata Ontology and Model

1
The ABC Metadata Ontology and Model

• Carl Lagoze, Cornell University
• Jane Hunter, DSTC

2
Some URLs

• This talk
• http//metadata.net/harmony/dc2001.ppt
• Latest paper version (JODI submission)
• http//metadata.net/harmony/JODI_Final.pdf
• Harmony project
• http//metadata.net/harmony

3
Harmony Project

• DSTC/JISC/NSF funding
• Assumption
• There is no lingua franca for metadata (Warwick
  Framework to RDF and XML namespaces)
• Although things like DC may provide basis for
  pidgin exchange
• Goal
• Investigate common principles and concepts
  underlying multiple descriptive models

4
ABC Goals

• Conceptual basis for understanding and analyzing
  existing metadata vocabularies and models
• Provide conceptual guidance to communities
  developing metadata models
• Provide a basis for mapping among vocabularies
  and models

5
Roots and Collaborations

• DCMI
• MPEG-7
• CIDOC/CRM
• INDECS

• CIMI
• RDF
• DAML
• Conceptual Graphs
• IFLA FRBR

6
ABC Model Overview

• (Digital) objects have inherent lifecycle
  characteristics
• Model creation, evolution, and transformation of
  objects over time.
• Notions of temporality are given first-class
  status
• Measuring utility of metadata through
  query-ability
• Ability to answer who, what, when, where comments
  that are difficult in simpler models

7
(No Transcript)
8
(No Transcript)
9
Entities and their Properties

• Entities are anchor points for set of properties
• Entities "change" by modification of property
  sets
• Model does not distinguish between "change of
  nature" and "change of description"
• Dual facets of Entities
• Universal object and its property set that is
  "global" to the description
• Existential object instances and their property
  sets that are periodic ("stateful")

10
Situations

• Establish a time period
• granularity determined by the longevity of entity
  state within situation
• Situations and actualities
• Situations provide context for associating
  entities in their existential facet.
• Entities can exist out of situation to express
  their universal properties

11
Events

• Transition marker point between situations
• Always have a time property
• Levels of increasing knowledge
• Something happened at a time (that caused change
  in situation)
• Type thing happened at some time
• Anchor point for multiple actions (verbs) within
  a happening.

12
Events, Actions, Agents

• Events provide a context for associating actions
  ("verbs"), 1-n
• Actions provide a context for participation of
  agents, 1-n
• Participation type can be specialized for domain

13
Causality verbs predicates

• Association of actions to Actualities in
  preceding situation is weak
• Increasing knowledge of association (esp.
  important in rights management)
• involves
• hasTool
• hasPatient

14
Intellectual Property

• Notion of "ability to copy" is the determining
  factor
• Promote abstraction to work and actuality to
  manifestation and item

15
Querying instances of the model

• Goal ability to ask questions like show me all
  the translations of War and Peace between 1980
  and 1990
• Squish query language SQL like
• http//swordfish.rdfweb.org/rdfquery/
• RDQL plug in of query mechanism to RDF API
• http//www-uk.hpl.hp.com/people/afs/RDQL/
• DSTC XML query interface
• http//www.metadata.net/harmony/Demos.htm

16
Future Work

• ABC API built on top of Jena
• http//www.hpl.hp.com/semweb/jena-top.html
• Workflow modeling tracking provenance
• Story telling description user interface

17
Asking the big questions

• How much metadata is needed?
• What is the use of metadata (search engine
  architecture query issues)?
• What are the tradeoffs between human effort and
  automatic generation (economic issues)?
• What are proper models for the proper contexts?