A review of metadata: a survey of current resource description formats
Work Package 3 of Telematics for Research project DESIRE (RE 1004)
Table of Contents
The Government Information Locator Service (GILS) has a draft application profile (see <URL:http://www.usgs.gov/public/gils/prof_v2.html>) that is intended to be submitted to the Open Systems Environment Implementors Workshop/Special Interest Group on Library Applications (OIW/SIG-LA). It has already had an Implementors Agreement approved by the OIW in May, 1994. The Federal Information Processing Standard Publication (FIPS PUB) 192 (see <URL:http://www.ncsl.nist.gov/fips/fips192.wp>) references the application profile. This specifies the GILS attribute set for Z39.50 for GILS servers and clients that support Z39.50 Version 2.
The GILS was setup by the US Federal Government in order to provide the general public and its own employees with a means for locating useful information generated by the many government agencies. As such its constituency of use is very broad; literally anyone is likely to be able to search for resources using GILS and many different agencies are likely to use a variety of staff to generate their part of the overall GILS framework. Originally GILS was intended to force each agency to provide a set of locators that "together cover all of its information dissemination products" (Executive Office of the President, Office and Management and Budget, OMB Bulletin, no. 95-01, Dec. 7, 1994 <URL:http://www.usgs.gov/gils/omb95-01.html>). However in reality some agencies are using GILS as generic metadata records for many resources and others are hardly using it at all.
GILS is a fairly complex metadata format, partly because of its breadth of coverage and partly because its design has been heavily influenced by the MARC and Z39.50 communities. Although it is possible that simple GILS records could be created by untrained staff, the format permits very rich and complex records to be created. As mappings to/from USMARC are provided in the GILS documentation it seems sensible to assume that at least some of the GILS records will be derived from USMARC records, which are themselves quite complex to create correctly.
The GILS concept builds upon many international standards and has resulted in the creation of the GILS profile for Z39.50 servers and clients. This is gaining some support, mainly as a result of pressure from the US government, but is not nearly as widely implemented and deployed within the Z39.50 community as say BIB-1. Indeed, many of the targets linked to from the GILS information pages (see <URL:http://www.usgs.gov/public/gils/targets.html>) are just straight library catalogues with Z39.50 servers returning only MARC or SUTRS records and aren't even run by a part of the US Federal government. In the GILS community these servers are said to be providing "less than full GILS functionality." (<URL:ftp://ftp.cni.org/pub/forums/gils/log9604>)
A number of other governments, such as the Canadian and Australian governments, are looking at the work done by the US GILS programme. Whether these are adopted on a large scale and what importance they will have in the future information society remains to be seen.
The information for this section is extracted from Annex E of the second draft of the GILS Application Profile. (Application profile for the Government Information Locator Service, Draft version 2. <URL:http://www.usgs.gov/gils/prof_v2.html>) This defines all of the elements in the GILS Core Element Set. These elements are defined as either being repeatable or not repeatable; the repeatable elements may appear more than once in a single GILS record whereas the not repeatable elements can only appear zero or one times.
Some of the elements are constructed from two or more subelements. For example, the Controlled Subject Index element is a grouping of subelements for Subject Thesaurus and Subject Terms Controlled. The grouping can be nested and is in this case; Subject Terms Controlled itself is a group formed from a repeatable subelement called Controlled Term.
The basic decriptive (bibliographic) elements included in the GILS Core Element set:
The subject description elements included in the GILS Data Element set are:
GILS Data Element records use the Availability Linkage and Availability Linkage Type data elements to specify the URI and MIME type respectively of the resource that the record is pointing at. These fields may be repeated within a single GILS record. There is also a set of Cross Reference elements that are used to refer to other, related GILS records. This set contains the Cross Reference Linkage and Cross Reference Linkage Type elements that indicate the URI and MIME type of the related record. The Cross Reference elements can also be repeated.
The resource format and technical characteristics and prerequisites are detailed in the following GILS Data Elements:
GILS Data Elements contain a number of fields to provide contact information. These include:
The administrative metadata required to maintain a GILS record is held in the following GILS Data Elements:
The GILS Core Data Element set provides the following elements for dealing with issues of provenance and record/resource data source.
Terms of availability and legal restrictions on records and resources (including but not limited to copyright) are included in the following elements from the GILS Core Data Element set:
The definitions of the GILS Data Elements is given in Annex E (GILS Core Elements) of the GILS Application Profile. There is also a U.S. National Archives and Records Administration publication called "Guidelines for the Preparation of GILS Core Entries" (<URL:http://www.dtic.mil/gils/documents/naradoc/>). This is intended to specify which elements are mandatory in specific contexts within the US Federal Government, and also gives examples of customary usage of specific elements.
It should be noted that it is also permissible to use locally defined elements within GILS records in addition to the GILS Core Element set. Some of these elements may themselves be well known elements in other Z39.50 application profiles or other information systems.
GILS Data Elements are available in an extended attribute-value pair format and the GILS Application Profile also provides a mapping to and from USMARC Tags and GRS-1 record syntax.
For physical transfer a GILS record may be delivered as a USMARC, GRS-1 or SUTRS record according to the GILS Application Profile. There may of course be further transfer encodings applied to these basic formats to allow the records to be sent through hostile environments. For SUTRS records, the GILS documentation defines an explicit preferred ordering to the output of the attributes so that the elements near to the top of the record are those most likely to show whether the record is useful to a searcher. However servers and clients are free to provide other orderings in addition to the preferred ordering.
The GILS Data Element set contains an element called Language of Resource that indicate the language of the resource that the record points at. There is also a Language of Record element that specifies the language that the GILS record itself is written in. Both of these are in the USMARC three character alpha code.
The Cross Reference elements of the GILS Element Set provides for the ability to describe relationships between records. The Cross Reference element subsets are also intended to be used inside Controlled Subject Index Subject Thesaurus structures to describe where to acquire and reference the thesaurus.
GILS is best described as being fairly high on the scale of fullness and complexity. For example in addition to all the elements described above for dealing with Document Like Objects (DLOs), it also contains a number of elements subsets for dealing with simple geospatial and temporal metadata. However it does not offer the range of specialised metadata formats that some of the more advanced geospatial applications require, such as percentage cloud cover.
GILS servers are often implemented using Z39.50 servers, although this does not appear to be mandated absolutely as some US Government institutions are providing their GILS records by other means such as WAIS and HTTP. Some agencies are using the GILS records to generate HTML documents suitable for browsing by WWW browsers and some have either provided CGI front ends to their Z39.50 servers or loaded the records into another web accessible database in order to allow users to search their resources using a normal WWW browser.
Most US Federal Government agencies now have GILS records deployed, and adoption of this format is being investigated by several other governments. Funding and encouragement from the US Government is also causing several companies, such as AOL and WAIS Inc, to start developing GILS compliant Z39.50 servers. Some of these will be freely available, whilst others will be commercial products.
|Next||Table of Contents|