Slides used for presentation at MPEG-7 Seminar, 9th April 1997

Linking text and image retrieval

Martin Bryan

SC18 WG8 - Document Description and Processing Languages

mtbryan@sgml.u-net.com

Existing SC18 WG8 Standards

• ISO 8879:1986 Standard Generalized Markup Language (SGML)
• ISO 9541:1992 Font Information Interchange
• ISO 10179:1996 Document Style Semantics and Specification Language (DSSSL)
• ISO 10180: 1996 Standard Page Description Language (SPDL)
• ISO 10744:1992 Hypermedia/Time-based Structuring Langauge (HyTime)

Planned SC18 WG8 Standards

• SGML Extended Facilities
• • includes Formal System identifiers
• ISO-HTML
• Standard Music Description Language (SMDL)
• Standard Multimedia Scripting Language (SMSL)
• Topic Navigation Maps

HTML - the language of the Web

• Formally defined using SGML
• Predefined set of markup tags
• Some tags have pre-defined behaviour
• • link anchors
  • images and image maps
  • forms
  • Java applets
  • scripts and styles
• Presentation format browser dependent

XML - SGML for the Web

• Well-formed subset of SGML suitable for incremental browsing
• All elements are fully specified
• No missing tags or attribute values
• Fixed 10646-based SGML declaration
• Notations can be user-defined
• • But no control via data attributes
• Styles controlled using DSSSL

SGML - the generalized solution

• Base language for HTML, XML, HyTime, SMDL, SMSL and Topic Navigation Maps
• No pre-defined elements or attributes
• Users must specify formal Document Type Definition (DTD)
• No default notations for images
• • ISO TR 9573-9 identifies standardized notations

Embedding non-SGML data

• Need notation processor for each notation identified in DTD
• Non-SGML data stored in uniquely named external storage location
• Entities can be assigned data attributes
• • Data attributes are declared as properties of specific notations
  • Data attributes are entered as part of an entity declaration

DSSSL - complex formatting

• Complex style semantics clearly defined
• • Extensible set of measurement units
  • DSSSL-O subset for on-line applications
• Scheme-based syntax for specification
• Tree transformation option allows documents to be merged, split and/or reordered
• SGML Document Query Language (SDQL)

HyTime location addressing

• Name locations
• • IDs, entity names, alternative name spaces
• Coordinate locations
• • finite co-ordinate spaces (for static and moving images), tree, path, relative, schedules and lists
• Semantic locations
• • data, SDQL query or manual bibliographic

HyTime links

• Can be embedded or externally stored
• Different types based on HyLink
• • clink, ilink, varlink, agglink
• Can link to multiple points and to spans
• Assign anchor roles to location lists
• Can control link and list traversal
• Can define different rules for each anchor

Topic Navigation Maps

Thesauri

Topic relationships

SC18 WG8 and MPEG-7

• Linking text navigation to image searches
• Adding image descriptions to topic navigation maps
• Adding music descriptions to sound via SMDL
• Creating integrated resource knowledge-bases

Sources of further information

• European Commission Open Information Interchange initiative
• • OII Standards and Specification List (http://www.echo.lu/oii/en/oiistand.html)
  • Monthly Report on Multimedia and Hypermedia Standards Activity (http://www.echo.lu/oii/en/activity.html)
• SC18 WG8 Web site (http://www.ornl.gov/sgml/WG8/wg8home.htm)
• Information on the development of the Topic Navigation Maps standard can be found through the home page of The SGML Centre (http://www.sgml.u-net.com)