The Fish-Search
is a robot, developed at the Eindhoven University of Technology.
It simulates the behaviour (and life cycle) of schools
of fish searching for food. The fish-search does not build an index-database
but searches for specific information (asked for by the user) on the fly.
- There are two versions of the Fish-Search: one that is integrated into
the browser (Tübingen Mosaic 2.4.2) and one that runs as a CGI-script
on a server.
- The search starts from a single (HTML) document, which should contain
interesting links to other documents.
- The query consists of a number of words or a regular expression.
Documents containing these words or matching the
expression are considered relevant.
- During the search a list is built of documents that (still) have to be
retrieved and checked for relevance.
- Links found in relevant documents are checked before links found
in irrelevant documents.
There is a Fish-Search Server at
http://www.win.tue.nl/cgi-bin/fish-search.