GILS Home | About | Technology | Standards | Policy | Index | Search |
On this Page:
At the level of standards and technology, GILS defines a certain set of Internet search requirements. Although it can be extended to serve various other information communities, the basic set is sufficient for many purposes.
The international standard for information search is ISO 23950 "Information Retrieval Application Service Definition and Protocol Specification for Open Systems Interconnection" (identical to the U.S. national standard ANSI/NISO Z39.50). The standard supports full-text search but also supports large, complex information collections.
The GILS standard is an international standard profile of ISO 23950. It specifies how to express a search and return results. It does not specify how network servers manage records or how clients use records.
GILS specifies an interface at the server side of a client-server interconnection over the Internet. At this interface, a GILS-compliant server interoperates in specific ways with client software A GILS-compliant server operates like a bibliographic catalog, but GILS uses only a few of the many optional specifications in the catalog standard. Most systems for searching bibliographic catalogs can search GILS servers and systems built to search GILS servers can search most bibliographic catalogs. Servers compliant with the ISO 23950 Geospatial Profile (GEO) or the Catalog Interoperability Proifle (CIP) are also compliant with the GILS standard.
The standard for searching bibliographic catalogs is one of several standards for network searching. Database searching is largely standardized on Structured Query Language (SQL). Directory search uses the Lightweight Directory Access Protocol (LDAP) standard. Other standards address searching document properties and content, including current work on the Extensible Markup Language (XML). Systems have demonstrated interoperability using GILS in combination with these and other network standards. GILS has also been used with Open Systems Interconnection and object-oriented architectures (e.g., CORBA) on the desktop and local networks, as well as directly over the Internet and with HyperText Transfer Protocol (HTTP). At present, it appears that the GILS standard search mechanism can be used in any operating environment and is well positioned to evolve with future network information discovery developments.
Client software to exploit GILS servers may be implemented within a single-user desktop application or as part of a gateway or other facility that handles interaction with multiple servers or protocols. The client-server interface specified in the GILS Profile is completely independent of any user interface and is therefore compatible with not only browsers but with other kinds of automated processing.
At the server side, the GILS interface is also independent of any specific database model or structure, and of the specifics of search engine implementations. This means that GILS is compatible with most databases, file systems, electronic mail, document management, and other methods of managing data, information, and knowledge.
The GILS interface is an Internet client-server protocol specification, as are the Hypertext Transfer Protocol (HTTP) and File Transfer Protocol (FTP). The GILS interface specifies how an Internet server supports a search service. In addition to some common Internet facilities, the GILS interface requires specific server responses for each of the three specified client request messages: Init, Search, and Present. The "Init" request/response messages communicate the protocol version and certain other session parameters. The "Search" request/response messages communicate a standard search vocabulary and syntax, including references to standardized search attributes with registered semantics. The "Present" request/response messages provide the machinery to deliver specific result records from the set of records that satisfied the search.
GILS-compliance requires the usual comparison operations ("less than", "greater than", "equal to") applied with dates, words, and phrases. It also requires the server to handle nine concepts: Title, Originator, Distributor, Record Source, Subject Terms--Controlled, Subject Terms--Uncontrolled, Date Last Modified, Any (i.e., "full-text"), and Local Number.
Following the principle of semantic mapping, the GILS interface provides a base of common search attributes that the public can use when searching. For any databases to support a GILS search interface, there needs to be a "map" relating these common characteristics to whatever fields are available in the database. For example, if a particular database has records each with the fields "Data Set Name" and "Owner", the database administrator might map the field "Data Set Name" to the search attribute "Title", and the field "Owner" to the search attribute "Originator".
The following semantics are registered in the ISO Basic Semantic Registry for the concepts used for search in the GILS Profile standard (these are known in ISO 23950 as "Use Attributes"):
Title: The name given as the distinctive designation of the information resource.
Local Number: The local number of the locator record of the information resource.
Originator: An organization or a group of persons that is identified by a particular name.
Date Last Modified: The date of last modification of the locator record of the information resource.
Distributor: The name of the person acting as distributor of the information resource product.
Subject Terms--Controlled: The controlled vocabulary associated with the subject of the information resource.
Subject Terms--Uncontrolled: Subjects headings defined locally.
Any: The record is selected if there exists a Use attribute that the target supports such that the record would be selected if the target were to substitute that attribute.
Anywhere: The record is selected if the term value (as qualified by the other attributes) occurs anywhere in the record.
West Bounding Coordinate: The value of the western- most coordinate of the location of the coverage of the information resource.
East Bounding Coordinate: The value of the eastern- most coordinate of the location of the coverage of the information resource.
North Bounding Coordinate: The value of the northern- most coordinate of the location of the coverage of the information resource.
South Bounding Coordinate: The value of the southern- most coordinate of the location of the coverage of the information resource.
The table below shows the combinations of attributes specified for a GILS-compliant server. GILS-compliant servers must respond to search requests that include three kinds of attributes: Use Attributes, Structure Attributes, and Relation Attributes. For example, to find documents with "USGS" in the Title, a search request would specify that the Use Attribute is "Title", the Structure Attribute is "Word" and the Relation Attribute is "Equal". At the protocol level, each attribute is identified by its assigned number as registered in the Attribute Set in force for the session. In addition to the Geospatial Profile, all attributes needed for GILS-compliance are registered in the Bib-1 Attribute Set and the GILS Attribute Set (both of these Attribute Sets are part of the ISO 23050 version 2 standard).
|Title||Word, Word List||Equal|
|Local Number||Word, Word List||Equal|
|Originator||Word, Word List||Equal|
|Date Last Modified||Date||Greater Than, Equal|
|Record Source||Word, Word List||Equal|
|Distributor||Word, Word List||Equal|
|Subject Terms-- Controlled||Word, Word List||Equal|
|Subject Terms-- Unontrolled||Word, Word List||Equal|
|Any||Word, Word List||Equal|
|Anywhere||Word, Word List||Equal|
|West Bounding Coordinate||Coordinate||Less Than, Less Than or Equal, Equal, Greater Than or Equal, Greater Than, Not Equal|
|East Bounding Coordinate||Coordinate||Less Than, Less Than or Equal, Equal, Greater Than or Equal, Greater Than, Not Equal|
|North Bounding Coordinate||Coordinate||Less Than, Less Than or Equal, Equal, Greater Than or Equal, Greater Than, Not Equal|
|South Bounding Coordinate||Coordinate||Less Than, Less Than or Equal, Equal, Greater Than or Equal, Greater Than, Not Equal|
|From a data store perspective, it is useful to think of the GILS interface as an additional gateway. On the Internet, this is like supporting FTP access to a collection of data in addition to supporting HTTP access. The GILS interface would typically occupy a different port on the same server, though it can also be located on another server if desired. The well-known port number for Z39.50/GILS/GEO/CIP is "210" and there can be any number of "databases" at the port.|
An administrator of a GILS server will need to decide at the semantic level how to map the search attributes available in GILS to the fields available in the database. Some of the fields may be obvious (e.g., Any/Anywhere should probably be targeted at any available text fields; Bounding Coordinates should be mapped to lat/long if available), while others may have no appropriate mapping at all. Another decision involves what sort of output is to be delivered when the searcher gets a "hit". There should be at least two levels of output--a "brief" record suitable for display as an item in a "hit list", and a "full" record to be delivered when the searcher selects that record from the hit list.
It is probably not a good idea for the administrator to make these decisions alone. The best approach is to define a simple mapping and see whether real searchers find it useful. Throughout the prototyping, it will be important to involve searchers outside of the usual primary audience. After all, the primary audience is likely already being served well by the interfaces specifically built for their needs. The GILS interface provides the opportunity to help "everyone else" find information.