DEVELOPING AN MPEG7 DESCRIPTIVE APPLICATION PROFILE FOR DIGITAL VIDEO with mapping to Dublin Core

1
DEVELOPING AN MPEG-7 DESCRIPTIVE APPLICATION
PROFILE FOR DIGITAL VIDEO --with mapping to
Dublin Core Participants ViDe Video Access
WG Co-Chairs Grace Agnew, Rutgers Dan
Kniesner, Oregon Health Sciences
University Rutgers Libraries Yang Yu, Database
Programmer Ruth Bogan, Mary Beth Weber,
Catalogers
2
Definition of Metadata

• Data about Data
• Data that describes, defines or manages data

Pure metadata has meaning only in relation to
the primary data that is being described.
3
Role of Metadata

• Facilitate organization, discovery,
  interpretation, preservation
• Enable implementer and end user to make sense of
  information appropriateness, context, relative
  value Shared understanding!

4
METADATA SCHEMA

• Creates standardized metadata
• Understandable by the archive and by the
  user--shared understanding.
• Shareable across repositories
• Can be mapped to other schema to repurpose the
  metadata.
• Maintained by a standards body for durability
  and consistency

5
METADATA SCHEMA COMPONENTS

• Data Element - Atomic Unit of Meaning- Community
  Defined
• Attribute - Refines, Extends, Interprets data
  element
• Value - Information unique to each data element
  instance
• Constraint - Order imposed on data element
  expression for consistency semantic viability
• Label - contextual instance of data element
  name. How the data element displays on the web
  for the end user.

6
METADATA RECORD
Data Elements
Populated with information (value)
According to rules (Schema)
Author
Agnew, Grace
VALUE
AACR2 Formatting RuleLast name, first name.
Data Element
7
SCHEMA VS. APPLICATION PROFILE

• Application Profiles
• Customize standardized metadata to support
  community needs.
• Utilize one or more schema, but often add the
  communitys own data elements
• Define constraints, such as mandatory,
  recommended or optional
• Develop controlled vocabularies, data types and
  formatting principles for information within each
  data element.

8
CUSTOMIZING METADATA

• Support for unique needs of users

• Adds to metadata universe of knowledge

• Supports unique commonalities among a
  distributed user base

-

• May not be developed robustly

• Interoperability may be compromised

9

• Description Needs for Digital Video and Audio
• Consecutive medium
• Contexual Metadata that changes over time as the
  information changes
• Media streamed in multiple formats (RealVideo,
  MPEG1, etc.) and stored on multiple publication
  formatstape, DVD, etc.
• Nontextual indexing (keyframe summaries, speech
  recognition, etc.) as important as textual
  indexing.

10

• Dublin Core
• Evolved with the web itself
• Encourage web authors to put structured
  information in ltmetagt tags
• Goals Simplicity and Interoperability
• Can be expressed as HTML or XML
• 15 optional, repeatable data elements
• Currently in version 1.1
• http//www.dublincore.org

11
Dublin Core
15 OPTIONAL, REPEATABLE ELEMENTS
From Description of Dublin Core
Elements http//purl.oclc.org/metadata/dublin_cor
e_elements
12
How Useful is Dublin Core?

• Provides a great deal of flexibility.
• Easy to learn.
• Ensures interoperability with other schemes.
• Good transport protocol when expressed as XML

-

• Lacks support for multiple formats
• Lacks support for seriality
• Technical description (formats, containers,
  extent, etc.) is weak and not standardized.
• Support for relationships (whole/part
  sibling/related) is very weak

13
Dublin Core

• Interoperability compromised because no consensus
  on valuese.g., how to fill the fields.

ViDe Tried!

• Issued Video AP in 2001
• Very functional but very complex.
• Violated the dumb down rule
• Decision Dublin Core is not rich and functional
  enough for DV

14
MPEG-7 Multimedia Content Description Interface

• Synchronization between content and description
• Supports description for intellectual content
  and all physical versions of that content
• Recursive and hierarchicalwhole/part (segment)
  description. Strong support for related works
• Native XML with some customized extensions

15
MPEG-7 Description Tools
Description Schemes (structure) and Descriptors
(features)
16
MPEG-7 - Metadata synched to media.MPEG7 AP for
the MPEG4 Metadata stream recently released
Start 0002001 End 0007327 Segment Title
introductory movement
Start 0007331 End 0102532 Segment Title
Main Theme
Start 0107031 End 01336 Segment Title
Oboe Solo
17
MPEG-7

• Non-textual indexing - melody and speech
  recognition, color, shape, scene changes, etc.
• Textual format/Binary Format completely
  equivalent. You can use any functionality in
  textual or nontextual form.

18
MPEG-7
19
MPEG-7 Multimedia Content Description Interface

• Does not support description of analog or textual
  resources
• High-level textual description of component
  parts (table of contents) does not exist.
• Some duplication of descriptive information
  across MPEG7 descriptive schemes
• Documentation, examples and widespread adoption
  as a descriptive metadata standard is weak.

20

• PROBLEM
• MPEG-7 highly functional but steep learning
  curve. Training, examples - nonexistent
• Dublin Core simple to employ (although not
  simple to employ well!). Widespread use, but not
  enough value for DV

21
The Structure of Information (IFLA)
Work
Distinct intellectual or artistic creation
Intellectual or artistic realization of a work
(interpretation)
Expression
Expression
Physical manifestation of an expression. May
differ in physical format, but not in content or
interpretation
Manifestation
Manifestation
Manifestation
Unique physical instance of a manifestation.
Item
22
WORK
A B S T R A C T I O N
GONE WITH THE WIND
Intellectual / artistic content
EXPRESSION
Novel
Movie
Script
Interpretation
70 MM Film
MANIFESTATION
Paper
Physical recording of content
35 MM Film
PDF
DVD
HTML
MPEG2
ITEM
Copy in Blockbuster, Atlanta, GA
Single physical representation of a recording
24 Reels of film, MGM Archive
23
Dublin Core vs. MPEG7 The Challenges

• MPEG7 is a structured, hierarchical schema.
• Work described in CreationInformation DS
• Manifestation/Item described in
  MediaInformation and UsageInformation DSs

• Dublin Core is a flat schema that mixes work
  or intellectual content with single
  manifestation/item description
• (11 principle)

24
MANIFESTATION
CREATOR
TITLE
CreationInformation
SUBJECT
DATE, etc.
MediaProfile UsageAvailability
MediaProfile UsageAvailability
IDENTIFIER
IDENTIFIER
FORMAT
FORMAT
MediaInstance
RIGHTS
RIGHTS
25
MPEG-7
Work/Manifestation
Manifestation
Item
26

• MPEG-7 includes descriptive, technical and
  structural information
• Dublin Core Descriptive only, except for
  rights

• MPEG-7 includes meta metadata (metadata about
  the record itself)
• Dublin Core has administrative metadata (A-Core)
  in development, for linking to descriptive
  metadata

27
(No Transcript)
28
CREATOR

• Person or body primarily responsible for content
  of the resource.
• Non-specificity of Dublin Core problematic
• Need for meaningful roles
• issue of type--person, organization
• contact information

ViDe Decision Add the attribute Role to
Creator. Weakly supports the dumb down rule
29
CREATOR

• MPEG-7 Creator includes
• Role
• Agent DataType
• Agent DataType includes
• Sub-types PersonType PersonGroupType
  OrganizationType
• Can include contact information

30
CREATOR

• MPEG-7
• ltCreatorgt
• ltRolegt
• ltNamegtSpeakerlt/Namegt
• lt/Rolegt
• ltAgent xsitypePersonTypegt
• ltNamegt
• ltGivenNamegtGracelt/GivenNamegt
• ltFamilyNamegtAgnewlt/FamilyNamegt
• lt/Namegt
• lt/Agentgt
• lt/Creatorgt

31
CREATOR

• Dublin Core
• ltCreatorgt Grace Agnew lt/Creatorgt
• Or
• ltCreatorgt Agnew, Grace lt/Creatorgt

32
(No Transcript)
33
(No Transcript)
34
CREATOR

• DC Dumb Down
• ltCreatorgt Speaker lt/Creatorgt
• and
• ltCreatorgt Grace Agnew lt/Creatorgt

35
DATE

• Date associated with an event in the lifecycle
  of a resource.
• Issue in print world/legacy catalogs, date is
  used to filter for currency of information. In
  Dublin Core, Coverage is used instead.
• Issue Date and country of release are important
  concepts for moving images, particularly feature
  films.

36
DATE A COMPLEX MAPPING
MPEG-7
CreationInformation?CreationCoordinatesCreationDa
te CreationInformation?Classification?ReleaseDate
Dublin Core
Date.Created Date.Issued
Not AddressedDate of Digitization. Technical,
rather than Descriptive metadata.
37
Date of Creation Map
38
Date of Release / Issue Map
39
WE BROKE THE SCHEMA
40
SUBJECT
Dilemma MPEG-7 schema currently allows only one
subject element per description ltminoccurs0/gt
Answer We edited the schema for our database and
inserted a note explaining what we did lt!Edited
by Rutgers University Libraries to support
unlimited subject elements--!gt ltminoccurs0,
maxoccursunbounded/gt
41
AND THE SCHEMA BROKE US
42
MPEG-7 ISSUE DURATION
PROBLEM MPEG-7 does not include a content
description data element for Duration
ViDe Solution Duration is critical information
for sequential media. Use SegmentDS, even though
it is intended for Navigation Access
43
SegmentDS Included in Navigation and Access
intended for nontextual bitstream processing.
44
FORMAT Part of the complex mapping doc!
45

• Next Steps
• Release the database in May
• Cataloging utility for Moving Image
  Collectionslooking for small science DV
  collection!
• Refine, develop registry and users guide
• Develop music AP for Rutgers Institute for Jazz
  Studies collection
• Test in some Media Asset Management systems.

46
(No Transcript)
47
Validates all entries against the MPEG7 Schema
for export
48
MORE INFO
Websites http//www.vide.net http//gondolin.rut
gers.edu/MIC
People
gagnew_at_rci.rutgers.edu kniesner_at_ohsu.edu