| Provider |
Harvest was originally developed by the Internet
Research Taskforce Group on Resource Discovery (IRTF-RD) as part
of an ARPA-funded project (Bowman et al 1994). The final official
Harvest release (1.4pl2) forms the basis for current open source
development.
Download http://www.tardis.ed.ac.uk/harvest/
Community support: harvest-develop@tardis.ed.ac.uk mailing list
and comp.infosystems.harvest Usenet group |
|
Licensing requirements: Open source software
distributed under the terms of the GNU General Public
License. |
| Purpose |
Originally intended to be a generalised resource
discovery system built in a modularised fashion, designed as an
open-ended system using network protocols, programming APIs and a
standard metadata format (SOIF). |
| Approach |
A Gatherer component performs indexing of
arbitrary content, using the Essence system from the University
of Arizona (included in the package). Metadata records (in SOIF
format) are produced by summarisers; the metadata records are
made available by the Gatherer component; a Broker component
periodically fetches SOIF records from Gatherers and indexes
them. The Broker is searched either using the Harvest Broker
protocol, or by plugging in search engines. The package tries to
implement the modular design by reusing readily available
software (such as existing indexers e.g. glimpse, the Essence
system, existing HTTP servers), and Harvest is essentially the
plumbing to connect these components together over a
network. |
| Platform and
requirements |
Http server, Perl, C compiler. Runs on any modern
Unix-like system. |
| Components and their
technology |
|
| Server |
The Harvest Broker component serves this
role. |
| Database |
|
| Search
tools/clients |
User interacts via HTTP. Search is performed via
a web form. A Perl CGI script (nph-search) speaks the Harvest
Broker protocol to the Broker and returns HTML to the end
user. |
| API |
|
| Metadata Formats |
Only the SOIF metadata format (originally
developed for Harvest) is supported. |
| Extras/Other
features |
Very limited web-based administration features
are provided, e.g. one can initiate a gathering operation. |
|
A new version of Harvest is written entirely in
Perl. |
|
See review
in the Renardus report for additional comments. |