Metadata Formats Work Package 1 of Telematics for Libraries project BIBLINK (LB 4034)

Title page Table of Contents Previous - Next

5. Typology of Metadata

It is clear that the various communities involved in creating and using different metadata formats are strongly attached to their own formats. This is understandable if one keeps in mind such factors as the effort involved in reaching consensus on formats, the skill levels required to apply formats in a consistent way, and not least the heavy investment in existing systems. For these reasons alone it is unlikely that any one format will become dominant. It would also seem undesirable as the existence of a variety of formats allows for choice of the optimum format for use in a particular context.

The provision and exchange of trade and bibliographic data relating to publications has been part of the book world for a considerable time, and has increased with the availability of data in electronic form. Records are created at various stages and places in the process of supplying documents to the reader, they are created to serve different requirements but have overlapping functions. The Newbury seminar held in 1987 (Bibliographic records in the book world: needs and capabilities) considered whether there could be improvement in the 'information flow' whereby the most appropriate record content and format could be sustained throughout the process to meet the various users' requirements (publishers, suppliers, libraries).

In the book world there is still only the beginnings of a more organised evolution of the bibliographic record. The issue of 'a single record system' remains outstanding; at present there is limited integration of records used in the publishing/supplier world and the national bibliography. For electronic publications we need to consider whether it is an ambition to more fully integrate record supply, how far various needs and requirements can be met by the single record system, or whether we accept a more evolutionary system with different record formats available to various users to meet their different requirements.

5.1 Continuum of Complexity and Richness

A variety of formats have been placed in this table, positioned along a continuum from simple records (Band One) to complex, rich records (Band Four). The variety of record types identified in the bibliographic control process can be placed on this continuum as shown below.

Band One	Band Two	Band Three	Band Four
Proprietary simple records:	Dublin Core	MARC	ICPSR
NetFirst	IAFA	TEI independent headers	FGDC
AltaVista	RFC 1807		CIMI
Infoseek	SOIF		EAD
Publishers' CIP forms	CIP MARC	EDI messages
		SGML article headers

It is possible to extend this model to associate other factors with the position of the format on the continuum. The simplest record formats are used to create relatively unstructured indexes for locating items whereas the most complex records can be used as the basis of sophisticated analysis and navigational tools. Records can be associated with more or less 'rich' retrieval and analysis processes (Z39.50, emerging query routing, text analysis). The bands of records typically have common characteristics in other aspects, for example:

• creation method
  how the record is created whether by hand or automatically. This in turn will affect level an type of resource required, and the overall cost of record creation.

• function of record
  whether the record is used for location, selection, evaluation, analysis

• complexity of designation
  simpler records do not permit complex designation of sub-fields, qualifiers etc

• associated protocols
  more complex formats are associated with the more heavy-weight search and retrieve protocol (Z39.50) whereas the simpler formats tend to be associated with directory service protocols.

This pattern of association is summarised in the following table:

Simple			Rich
Location	Selection	Evaluation	Analysis
Robot generated	Robot plus manual input	Manually input	High level of manual input
Unstructured	Attribute value pairs	Subfields, qualifiers	Highly structured mark up
http with CGI form interface	directory service protocols (whois++) with query routing (Common Indexing Protocol)	Z39.50	Z39.50 (in future with collection navigation)
proprietary	emerging standards	generic standards used in information world	standards used in specialist subject domains

Within the context of BIBLINK we need to consider which Band of record is appropriate for further consideration. There may well be a different answer depending whether the requirement is for a CIP type record or a more detailed record.

5.2 Further Scoping Required

• what functions will the record fulfil?
  
• can we assume that we are not looking for proprietary solutions, but rather for standards supported by international (or at least European) consensus? If so then we can rule out Band 1 formats.
  
• is the aim to choose metadata format(s) which can also be used for print material?
  
• what will be the creation method for the records? How much manual effort will be available? Are there cost criteria?
  
• which Internet protocols will be used? Are there constraints imposed by the technical framework? (This will be handled in WP5)
  
• what level of designation is required? Will mapping to other formats be necessary and if so how much data loss during conversion is acceptable?
  

Project partners need to consider these issues and refine the scoping of the project to allow criteria to be drawn up for decision making.

While accepting the need for consensus from partners on these issues, we will assume that previous practice and discussion of these issues will inform future decisions. So for example we assume that there will be constraints on cost and staff available from national libraries and publishers; that we are attempting to identify standard formats that are controlled by authoritative agencies; and that the level of service provided by the national bibliographic agencies will be comparable to that provided for print material (whether this is at a CIP level or at a level consistent with the full record in the national bibliography).

Given these assumptions, the formats from Band One would be rejected as proprietary solutions. The formats in Band Four would also be rejected as too detailed for the service level required, too specialised in nature for general use, and too costly for system maintenance particularly in terms of specialist staff required with skill levels to manipulate and interpret such records.

Predicated on these assumptions as to scope, it is recommended that BIBLINK should concentrate on formats in Bands Two and Three for the exchange format. This does not preclude the possibility that conversion will be required from formats outside these Bands e.g. from more complex formats in Band Four into a simpler format, but that the formats for data exchange would be located in Bands Two or Three.

5.3 Topology of Metadata

One example of this is given by Bearman who proposes a reference model for business acceptable communication. (Bearman, David and Sochats, Ken. Metadata Requirements for evidence. Available at <URL: http://www.lis.pitt.edu/~nhprc/model.htm>). This defines clusters of data elements which would be required to fulfil the range of functions of a record. The functions of records are identified as the provision access and use rights management, networked information discovery and retrieval, registration of intellectual property, authenticity. The clusters of data elements are defined in six layers:

• Handle Layer
  
  • registration metadata or properties
    
  • record identifier
    
  • information discovery and retrieval
    
• Terms and Conditions layer
  
  • rights status metadata
    
  • access metadata
    
  • use metadata
    
  • retention metadata
    
• Structural Layer
  
  • file identification
    
  • file encoding metadata
    
  • file rendering metadata
    
  • record rendering metadata
    
  • content structure metadata
    
  • source metadata
    
• Contextual Layer
  
  • transaction content
    
  • responsibility
    
  • business function
    
• Content Layer
  
  • content description
    
• Use history layer
  

Within the UK, the BIC/BNBRF Book Product Information project has had as its main aim the identification of the content required for an EDI message to communicate product information through the book sector supply chain for non-serial items. This work has now developed into compiling exhaustive sets of data elements which might be used in this context.

• Line item identity and attributes
  
  • identification
    
  • type
    
  • tradability
    
  • relationships to other line item
    
  • publisher
    
• Work identity and attributes
  
  • Title data
    
  • Personal authors, data providers
    
  • corporate authors
    
  • description
    
  • subject
    
  • relationships to works
    
• Media specific attributes of a work
  
  • cartographic scale
    
  • video duration
    
• Attributes of a piece
  
  • media codes
    
  • dimensions
    
  • number of pages
    
  • supply medium
    
• Attributes relating to terms of trade
  
  • market area
    
  • publication date
    
  • terms of sale
    
  • price
    
• Attributes relating to sales promotion
  
  • Promotional text
    
  • Promotional images
    
  • Author promotion
    
• Message function
  
  • message type (replacement, update etc.)
    
  • effective date etc.
    

Line item:	the entity described in the Book Product message, in effect a tradable item
Work:	body of literary or intellectual content
Piece:	single indivisible physical published item.

Turning to the library world, IFLA has recently issued a draft for comment of its study of the functional requirements of bibliographic records. This study attempts to identify and define objects of interest (or entities) for users of bibliographic records i.e. what information the user expects to find in the record, and how that is used. As regards the intellectual content described by a particular bibliographic record, can be viewed at different levels, each of which can be related to different information seeking behaviours from the user. These entities are:

Work:	distinct intellectual or artistic creation
Expression:	realisation of the work in text, sound, music, image etc
Manifestation:	physical embodiment of an expression in book, sound recording etc
Item:	single example of a manifestation

Within the context of BIBLINK it would seem there needs to be a shared view among publishers and national libraries as to what 'entities' or 'objects' are of interest. Traditionally libraries have dealt with printed material to a large extent at the manifestation level e.g. particular editions of a book. As electronic formats offer increasingly varied manifestations, then creation and re-use of metadata relating to the 'work' will take on more emphasis. At a different 'entity' level, it may become desirable to describe collections rather than individual items e.g. for web-sites containing varied and changing information. The problem of the identification of different manifestations and expressions is perhaps more familiar to publishers who are accustomed to dealing with compilations, re-prints, new editions and other 'bundling' of works. The on-going work by BIC to formulate a non-serial DTD has recognised this in the proposed layering of data elements.

Recommendation

As part of the consensus building process, publishers and national libraries identify the objects and relationships which need to be represented in metadata describing electronic resources.

Next	Table of Contents