Metadata: the `killer application' for digital broadcasting?

Tom Worthington FACS

Visiting Fellow, Department of Computer Science, Australian National University and Director, Tomw Communications Pty Ltd

For the Australian Broadcasting Authority 2002 Conference "What Will Australian Audiences Want?", 29-30 April 2002, Canberra.

This document is Version 3.3 20 March 2002: http://www.tomw.net.au/2002/mka.html

Contents

  1. Abstract
  2. Introduction
  3. TV Anytime Forum's Metadata Specification
  4. The `killer application': custom programs on demand
  5. Conclusion
  6. References

Abstract

Existing digital broadcasting has a failed business model, as it ignores what the audience wants most: control. Metadata has provided a hidden "killer application" on the Internet and could be used to transform digital broadcasting into a viable service. A brief overview of metadata is provided. The latest metadata developments and their applicability to digital broadcasting are discussed.

Introduction

The broadcasting industry is at the point with digital broadcasting the telecommunications industry was with digital communications in 1994. A successful direct translation is being made from analogue sound and video broadcasting to their digital equivalents. There is an assumption that there will be a more advanced use of the the technology, but no clear direction as to its form, there is debate over standards for interactivity and worry about how to pay for the cost of the new technology.

The debate over telecommunications in the mid to late 1990's was around the "information superhighway". The Internet emerged from academic research relatively late in the debate and displaced what the telecommunications industry had envisaged as the superhighway. Current broadcasting debate is around "Interactive Television" (iTV), which is seen as an advanced development of digital broadcasting. Just as telecommunications industry attempted to adopt standards for the infobahn, but failed, it is likely that Internet derived technology will overtake "Interactive Television".

The TV broadcasting industry is undergoing a difficult period where it attempts to adopt interactive features for existing services. There has been an attempt, with limited success, an attempt to adopt some aspects of the Internet, with standards such as MHP (Worthington 2001a).

AUSTRALIA'S free-to-air television networks have agreed to use a new interactive standard, but they are far from unified about how to go about introducing it. The Nine Network has criticized the slow advancement of the interactive television standard, MHP, but insists it is still committed to the embryonic technology. Nine's director of digital services Kim Anderson expressed frustration with the development of MHP, or Multimedia Home Platform, yesterday, and said her network had proceeded with developments in the DVB-HTML standard in the meantime... Confusion over iTV standard Kate Mackenzie,The Australian, FEBRUARY 27, 2002

At the same time the broadcasting industry debates interactive TV, the issue of broadband access in Australia, which was relatively dormant for five years, has again become an issue. This follows the failure of Australian government policy which saw the rollout of two competing analogue pay TV networks. Many of the issues of the 1990's infobahn debate (Clarke & Worthington 1994) are again becoming prominent: what level of service is it reasonable to aim for in Australia?; who will pay for it? Telecommunication suppliers argue that consumers need some compelling content before it will be worth investing in a network to carry it, content creators argue that it is not worth creating content for a non-existent network:

TELSTRA says the federal Government should consider subsidies or tax incentives to help people buy the decoders needed to view digital TV, while again calling for new TV licences to be issued... "No commercial model other than a commercial free-to-air broadcasting licence will be able to accommodate the transmission costs," it said. "Telstra sees the cost of transmission infrastructure as prohibitive as it is fixed regardless of the size of the audience." Tax break for digital TV decoders Jane Schulze, The Australian, FEBRUARY 01, 2002

Those companies which have attempted iTV, have not found it a commercial success. Optus TV executives were talking up expanded interactive TV services:

OPTUS looks likely to make a commercial launch of its interactive television product, after it said a commercial iTV trial in Sydney would reach its take-up target by the end of next month. Optus consumer and multimedia marketing director Scott Lorson told a digital media conference in Sydney today that it will make its target for interactive TV take-up of 3000 subscribers by the end of March, with more than 2000 paying customers since the commercial trial launched in December... From: Optus iTV all set to go Kate Mackenzie, The Australian, FEBRUARY 28, 2002

A few days later they were bought off by the analogue Foxtel rival:

FOXTEL is poised to win the battle for the wallets of Australia's 5 million pay-TV viewers, emerging as the dominant force in an industry that has lost up to $4 billion in the past eight years. In a groundbreaking deal signed yesterday, Foxtel has bought off its main competitor, Optus TV, by assuming payments to Hollywood studios that, according to Optus executives, total more than $600 million over eight years. Foxtel takes pay-TV by throat By Geoff Elliott and Jane Schulze March 06, 2002

The enhancements to digital TV offered by interactivity include: "video-on-demand" (VOD) or near-video-on-demand (NVOD), electronic program guide (EPG), games, email, and other internet-like services. However, will consumers pay for these extra services? Will pay TV or free-to-air broadcasters be able to provide enough services to make it economic?

While wanting consumers to pay for services, the broadcasting industry wants to retain a similar level of control of the service as is done with analogue broadcasting. Ideally a broadcaster wants the viewer to see exactly what is offered, when it is offered, in the format it is offered. Interactivity is provided within very narrow limits.

Many of the issues surrounding iTV are to do with precisely defining a set of fixed Internet-like standards. Broadcasters argue that these need to be fixed so that consumers will receive a reliable quality service. Similar arguments were put in 1994 to say that the Internet was not acceptable as a consumer product as it was unreliable and variable.

Consumers voted with their feet and adopted the Internet, because it provided more control. Hardware, software, networks and content came from many different sources. There were, and are, technical complexities, network failures, commercial failures, but consumers still used the Internet.

The interactive TV conundrum can be similarly sidestepped by adopting a creative and flexible approach. This approach is already evolving with set-top-box manufacturers adopting the most popular parts of proposed iTV standards.

What is now required is the multimedia equivalent of the first graphical web browsers: something which can operate on existing networks, with a relatively crude service, but with compelling content and room for expansion. Existing internet technology can be grown to provide content which:

The first step is to describe the content, using metadata.

TV Anytime Forum's Metadata Specification

The TV Anytime Forum develops specifications for services based on consumer digital storage devices, such as the TiVo (who are one of the forum members). Australian forum members include Nine Network Australia. The forum will meet in Sydney in November 2002. Curiously the TV Anytime web pages are marked with a BBC R&D Copyright notice.

The Forum has Working Groups on: Business Models, Content Referencing and Location, Metadata, Rights Management and Protection, System and Transport. Metadata will be discussed in detail, but the Business Models work is unusual (and commendable) in a technical standards body:

The forum has working groups on Business Models, Content Referencing and Location, Metadata, Rights Management and Protection, System and Transport. The work of the Metadata group is discussed in detail below. The work of the Business Models group is also notable:

Models for key stakeholders such as consumers, content and service providers, advertisers & equipment manufacturers are considered. Recently they have included personalisation, targeting, segmentation and networking alongside basic 'present day' PDR functionality.

From: WG Business Models (BM), 01 February 2002

They have released draft Metadata Specifications:

Within the TV-Anytime environment, the most visible parts of metadata are the attractors/descriptors used e.g. in Electronic Program Guides (EPG), or in Web pages to describe content. This is the information that the consumer, or intelligent agents, will use to search and select content available from a variety of internal and external sources.

Another important set of metadata consists of describing user preferences, representing user consumption habits, and defining other information (e.g. demographics models) for targeting a specific audience.

The TV-Anytime Metadata Specification also allows describing segmented content. Segmentation Metadata is used to edit content for partial recording and non-linear viewing. In this case, metadata is used to navigate within a piece of segmented content.

From the TV-Anytime Metadata Specification working group, 2001

Describing TV Programs

An example of the structure the metadata standard is a description of the topic:

  1. Content
    1. NON-FICTION
      1. News
        1. Daily news
        2. Special news/edition
        3. ...
      2. Philosophies of life
        1. Religious philosophies
          1. Buddhism
          2. ...
        2. Non-religious philosophies
          1. Communism
          2. ...
    2. SPORTS
      1. Athletics
        1. Field
        2. ...
      2. ...
    3. LEISURE/HOBBY
    4. FICTION
    5. AMUSEMENT
    6. MUSIC AND DANCE
      1. Classical music
        1. Early
        2. ...
      2. Background music
      3. ...
    7. INTERACTIVE GAMES ...

    Adapted from Specification Series: S-3 On: Metadata Corrigenda 1 to S-3 V1.1, The TV-Anytime Forum, December 21, 2001 in file: ftp://tva:tva@ftp.bbc.co.uk/pub/Specifications/COR1_SP003v11.zip

This ontology raises interesitng issues for cultural diversity, which might be usefully explored by others.

Specification Documents

The Metadata working group released a first requirements document in April 2000. Their latest draft documents are:

Documents are publicly available from the forum web site, as zipped MS-Word documents. As with MHP and TeleWeb, this makes the standards difficult to find, download and read (Worthington 2002). In contrast Internet and web standards are provided in an easy-to-use text and web formats.

TV-Anytime uses XML syntax for defining what metadata will be used (the schema) and then to actually encode the actual data (instances). The metadata structure is defined using the XML Schema based MPEG-7 Description Definition Language (ISO 2001). XML Schema is a recent web standard for defining sophisticated data structures (Worthington 2001b), which uses the XML syntax.

While XML has advantages, being widely supported and easy to generate, it is a verbose text format. TV-Anytime propose using binary encoding and compression of the XML for efficiency. However, the draft documents contain notes indicating that there is some disagreement as to the format. It might be simpler if plain text was initially used and compression added as an option, if needed later. Given that metadata would represent a small amount information, compared to video data, the complexity of compression may not be worthwhile.

An example of a metadata definition is:

<complexType name="EventInformationType">

<sequence>
<element name="PublishedTime" type="dateTime" minOccurs="0"/>
<element name="PublishedDuration" type="duration" minOccurs="0"/>
<element name="Live" type="tva:FlagType" minOccurs="0"/>
<element name="Repeat" type="tva:FlagType" minOccurs="0"/>
<element name="FirstShowing" type="tva:FlagType" minOccurs="0"/>
<element name="LastShowing" type="tva:FlagType" minOccurs="0"/>
<element name="Free" type="tva:FlagType" minOccurs="0"/>
</sequence>

</complexType>

From TV-Anytime Metadata Specification Version 1.2, WD498 Part-A.doc in ftp://tva:tva@ftp.bbc.co.uk/pub/Plenary/TV096r2.zip

This describes a broadcast "event" (TV program), including the time at which the program is advertised as starting, duration, if it is a live broadcast or repeat and if the program is free.

Similarly the format can be used to describe an on-demand program:

<complexType name="OnDemandProgramPublicationType">
<complexContent>

<extension base="tva:ProgramLocationType">
<sequence>
<element name="PublishedDuration" type="duration" minOccurs="0"/>
<element name="StartOfAvailability" type="dateTime" minOccurs="0"/>
<element name="EndOfAvailability" type="dateTime" minOccurs="0"/>
<element name="URL" type="anyURI"/>
<element name="FirstAvailability" type="tva:FlagType" minOccurs="0"/>
<element name="LastAvailability" type="tva:FlagType" minOccurs="0"/>
</sequence>
</extension>

</complexContent>
</complexType>

From TV-Anytime Metadata Specification Version 1.2, WD498 Part-A.doc in ftp://tva:tva@ftp.bbc.co.uk/pub/Plenary/TV096r2.zip

This is used to describe programs that can be shown on demand (not broadcast), with the when it will first be available and the URL where it can be obtained.

Targeting metadata

The program metadata described above is analogous to a traditional printed TV guide. Its use is relatively well understood and specified. In contrast the Targeting metadata, describes the intended audience for a particular content. This allows content to be selected by matching descriptions of the audience, rather than of the content. This is the newest part of the Metadata Specification (February 1, 2002) is the first draft and shows interesting possibilities beyond conventional TV guides:

... enables system operators e.g. program copyright owners, distributors and broadcasters, to influence or enforce the selection of content recorded or presented to a user e.g. at end-user equipment. This is going to be extended by defining additional targeting metadata (e.g. including demographics, consumption preferences, etc.). To be completed with a reference to the business model's phases and model when stabilised on targeting.

From Draft Section on targeting, February 1, 2002, WD499 Part A Targeting.doc in ftp://tva:tva@ftp.bbc.co.uk/pub/Plenary/TV120.zip

Targeting metadata could prove exciting for marketing, by automatically matching content to desired features of the audience. Examples used in the proposal include Car ownership, Foreign Holidays, Pet Ownership and demographics such as age, sex , socio-economic group and consumer preferences. Selecting content based on mood is also proposed. These features might appeal to advertisers, but it is not clear that mood altering programming would be accepted by viewers. However, they may be of value in assembling interactive programs from components, identified with Segmentation Metadata.

Segmentation Metadata

Segmentation Metadata probably provides the most scope for new forms of interactive content with TV-Anytime, but will cause the most problems for current conventional broadcasters. As well as being able to identify whole programs, segmentation allows segments within an audio visual stream to be identified by their start and stop time. This allows content to be restructured and re-used. Highlights can be presented, an index to sections of AV material or the material can be virtually re-editied. The Segmentation Metadata can be provided with the AV content or generated by the viewer or others. An example given by TV-Anytime is in repurposing of content for educational purposes.

<element name="SegmentInformation" type="tva:SegmentInformationType"/>
<complexType name="SegmentInformationType">

<sequence>
<element name="ProgramRef" type="tva:CRIDRefType" minOccurs="0"/>
<element name="Description" type="tva:BasicSegmentDescriptionType" minOccurs="0"/>
<element name="SegmentLocator" type="mpeg7:MediaTimeType"/>
<element name="KeyFrameLocator" type="mpeg7:MediaTimeType" minOccurs="0" maxOccurs="unbounded"/>
</sequence>

<attribute name="segmentId" type="ID" use="required"/>
<attribute name="version" type="integer" use="optional"/>

</complexType>

From: Metadata, The TV-Anytime Forum, August 17, 2001, SP003v11.doc, ftp://tva:tva@ftp.bbc.co.uk/pub/Specifications/SP003v11.zip

The Segment Information includes a reference to the program the segment belongs to, a description of the content, of the segment. location of the segment (start time and duration).

The `killer application': custom programs on demand

TV Anytime's metadata specification was defined with conventional linear TV programs in mind. However, provision for on-demand, targeting and segmentation allow for more creative possibilities.

Digital networks, such as Canberra's Transact, could provide a low cost platform for distribution of experimental work. Those formats which prove popular could then have a business model built for them and become commercial services. Filmmakers need not limit themselves to conventions linear video, but could use targeting and segmentation.

Where a wired digital network is not available, a wireless system could be used. The content could be automatically adapted for the lower bandwidth network and hand held display device (such as a PDA):

The Minister for Communications has ordered an investigation into the merits of wireless LANs as a possible solution to the 'last mile' problem. A spokesperson for Senator Richard Alston has confirmed the department is preparing a report into the potential of the international IEEE 802.11 wireless standard. The report will investigate the usefulness of the technology for regional and rural areas throughout Australia as well as SMEs and home users. Govt looks at wireless to solve 'last mile' issues, By Kelly Mills, Computerworld 26 February, 2002 8:15, Sydney, Australia

An additional costly and complex broadcasting infrastructure is not needed for delivering video content. The same model as used for creating and distributing web sites can be used: content is prepared on a personal computer and then uploaded to a server for delivery. Community programs, including news, could be created by individuals shooting material and uploading it from home to a server where others could select material for a program. The server and network can then deliver content on-demand or at preset times. If some content becomes very popular, then larger servers based on multiple PCs, similar to the Bunyip super-computer could be used (ANU 2001).

Multimedia is currently poorly served by video editing software, which is designed to simulate the analogue film editing process. It should be possible to produce hybrid software, which adds functions for the creation of storyboards and other multimedia tools to office automation software.

Open Office is a project to produce open-source office software project with word processing, spreadsheet and graphics functions. Open Office uses an XML based file formats and has the ability to convert to and from Microsoft office formats.It has been demonstrated that it is feasible to transform XML Open Office documents to different formats using short simple computer programs (Barnes 2002). It should be feasible to add additional XML formats, such as the TV-Anytime Metadata Specification, to Open Office format documents.

Integrated software would allow imported text to be marked up as a film script. The script could then have a storyboard added and this played back with text displayed on screen or spoken by a synthetic voice. The video could then be shot and automatically edited to a rough cut. This would allow multi-media work, with print, web, audio, video and interactive versions to produced in one work.

Conclusion

The Internet's approach of experimentation can be used to explore for new business models for broadcasting. Metadata can provide the "killer application" to transform digital broadcasting into a viable service: control over content by the viewer. High speed wired networks and medium speed wireless ones could provide the platform for distribution of experimental work, in much the same way the web started. Content servers based on personal computer technology can provide multimedia on demand to the community. Hybrid open source software, using XML can add multimedia to office applications and bring content development to the general community.

References

Personal profile

Tom Worthington Tom Worthington is a Visiting Fellow in the Department of Computer Science, Faculty of Engineering and Information Technology at the Australian National University. He is an electronic business consultant, author of the book Net Traveller and information technology professional, with 17 years experience.

Request for Comments

The text of this presentation will be refined here. Comments, corrections and suggestions would be welcome.

Further Information

Comments and corrections to: webmaster@tomw.net.au

Copyright Tom Worthington. 2002