8.6 Summary and Conclusions
the set of values encoded in a typical header file,
an Import service must be broader than simply
reading the header; for example, some other mechanisms must be used to obtain values for characteristics not defined within the header.
As conceived in the early 1990s, before Web
tools dominated the Internet-based dissemination
of information, EIS initially used basic facilities
analogous to anonymous file transfer protocol
(FTP) to support information exchange between
designated EIS sites. However, by the time the implementation was actually stable enough for public distribution, it was clear that EIS would meet
the fate shared by other niche data dissemination
systems: no matter what advantages these systems
might offer, the explosive growth in the popularity
and ease of use of Web-based tools rapidly led the
target user community to expect that information
should be disseminated over the Internet with Webbased tools, not other customized software. As
such, EIS development parallels other early efforts
in building network-based data dissemination systems that were caught short by the rapid emergence
of Web-based tools as de facto standards. Simply
put, users quickly began to expect data to be found
through the use of standard Web browsers and
search engines, rather than with tools and interfaces
customized for any other special-purpose system.
As a result, the designers of any custom network
data repository now face three choices: simply
abandon their efforts; continue to promote their
system as an alternative to Web-based access; or
reengineer their system to retain novel data classification and access components, but add some sort
of standard Web-based access mechanism as a front
end to the customized system. The last choice is
obviously the most desirable, because it salvages
the major part of the original effort. More importantly, this is also a way to provide access to information, such as the data repository's metadata,
that is otherwise lacking in simple collections of
Web pages or simple free text indexing. In the case
of EIS, we have taken this approach of reengineering the interface. The newest version of EIS
replaces the original, customized graphical user interface with a more familiar Web interface and uses
Java mechanisms to link customized EIS databases
and search mechanisms to standard Web-access
techniques. This assures that user queries can be
generated with familiar interfaces, yet be translated
on the fly into appropriate EIS-Web searches. We
also have been working to describe classification
hierarchies consistent with FGDC metadata standards and for which FGDC indexing tools can also
be used (Ford et aI., 1997). In this respect, current
117
EIS development attempts to complement, rather
than compete with, both the FGDC clearinghouse
program and the mass-market mechanisms in use
on the Web. This complementary approach uses
Web mechanisms for the basic interface associated
with query formulation and display. It also uses
Web-based protocols for information transfer.
However, it interjects customized processing elements between the Web interface and final processing so that W eb-formulated queries are directed
to specific metadata-based indices, rather than
generic Web indices. The use of specific metadata
attributes adds significantly to the organization of
the accessible information and provides considerable value to the searches that users formulate.
8.6 Summary and Conclusions
Our discussion traces the state of the art in data-set
dissemination from the pre-Internet days, dominated by local issues of data format and storage,
through early non-Web days, which saw the development of numerous customized access mechanisms
for electronic data repositories, to the current Webdominated environment. What has changed in recent
years is the sort of mechanisms through which
clients expect to be able to gain access to the information resources. What has not changed over
time is the need for repository managers to effectively organize their ever-growing information
resource collections. Thus the keys to maintaining
effective systems for data dissemination are the following:
1. To accept standard Web facilities as the basic
user interface and data-delivery element. It
seems futile at this point to invest resources in
offering software that competes with what has
emerged as the de facto standard.
2. To focus local efforts first on building rich,
content-specific descriptions of local repository
contents. The key is to move beyond simple sets
of Web pages into well-organized contentspecific catalogs or indices based on the metadata associated with repository components.
3. For organizations with no preexisting catalogs
and indices, to incorporate Web accessibility
from the beginning as these catalogs and indices
are built.
4. For organizations with significant existing investment in non-Web catalogs and indices, to
focus attention on linking existing indices to
Web-compatible forms, rather than converting
or abandoning local systems. This allows an or-
the set of values encoded in a typical header file,
an Import service must be broader than simply
reading the header; for example, some other mechanisms must be used to obtain values for characteristics not defined within the header.
As conceived in the early 1990s, before Web
tools dominated the Internet-based dissemination
of information, EIS initially used basic facilities
analogous to anonymous file transfer protocol
(FTP) to support information exchange between
designated EIS sites. However, by the time the implementation was actually stable enough for public distribution, it was clear that EIS would meet
the fate shared by other niche data dissemination
systems: no matter what advantages these systems
might offer, the explosive growth in the popularity
and ease of use of Web-based tools rapidly led the
target user community to expect that information
should be disseminated over the Internet with Webbased tools, not other customized software. As
such, EIS development parallels other early efforts
in building network-based data dissemination systems that were caught short by the rapid emergence
of Web-based tools as de facto standards. Simply
put, users quickly began to expect data to be found
through the use of standard Web browsers and
search engines, rather than with tools and interfaces
customized for any other special-purpose system.
As a result, the designers of any custom network
data repository now face three choices: simply
abandon their efforts; continue to promote their
system as an alternative to Web-based access; or
reengineer their system to retain novel data classification and access components, but add some sort
of standard Web-based access mechanism as a front
end to the customized system. The last choice is
obviously the most desirable, because it salvages
the major part of the original effort. More importantly, this is also a way to provide access to information, such as the data repository's metadata,
that is otherwise lacking in simple collections of
Web pages or simple free text indexing. In the case
of EIS, we have taken this approach of reengineering the interface. The newest version of EIS
replaces the original, customized graphical user interface with a more familiar Web interface and uses
Java mechanisms to link customized EIS databases
and search mechanisms to standard Web-access
techniques. This assures that user queries can be
generated with familiar interfaces, yet be translated
on the fly into appropriate EIS-Web searches. We
also have been working to describe classification
hierarchies consistent with FGDC metadata standards and for which FGDC indexing tools can also
be used (Ford et aI., 1997). In this respect, current
117
EIS development attempts to complement, rather
than compete with, both the FGDC clearinghouse
program and the mass-market mechanisms in use
on the Web. This complementary approach uses
Web mechanisms for the basic interface associated
with query formulation and display. It also uses
Web-based protocols for information transfer.
However, it interjects customized processing elements between the Web interface and final processing so that W eb-formulated queries are directed
to specific metadata-based indices, rather than
generic Web indices. The use of specific metadata
attributes adds significantly to the organization of
the accessible information and provides considerable value to the searches that users formulate.
8.6 Summary and Conclusions
Our discussion traces the state of the art in data-set
dissemination from the pre-Internet days, dominated by local issues of data format and storage,
through early non-Web days, which saw the development of numerous customized access mechanisms
for electronic data repositories, to the current Webdominated environment. What has changed in recent
years is the sort of mechanisms through which
clients expect to be able to gain access to the information resources. What has not changed over
time is the need for repository managers to effectively organize their ever-growing information
resource collections. Thus the keys to maintaining
effective systems for data dissemination are the following:
1. To accept standard Web facilities as the basic
user interface and data-delivery element. It
seems futile at this point to invest resources in
offering software that competes with what has
emerged as the de facto standard.
2. To focus local efforts first on building rich,
content-specific descriptions of local repository
contents. The key is to move beyond simple sets
of Web pages into well-organized contentspecific catalogs or indices based on the metadata associated with repository components.
3. For organizations with no preexisting catalogs
and indices, to incorporate Web accessibility
from the beginning as these catalogs and indices
are built.
4. For organizations with significant existing investment in non-Web catalogs and indices, to
focus attention on linking existing indices to
Web-compatible forms, rather than converting
or abandoning local systems. This allows an or-