Search tools like the Fish-Search, retrieve information from the
entire contents of (text) documents.
This enables the best judgement of whether documents are relevant for
the user's query or not.
The fish-search uses a simple algorithm to evaluate the relevance of
documents:
- The user can select whether all given words must occur or not.
(boolean and and or.)
- If some (or, if chosen, all) words occur in a document the document
is considered relevant.
- If more than one of the given words occur the document is more relevant.
- If the words occur often, relative to the size of the document,
the document is more relevant.
Information retrieval research shows that one should also
take into account the "average" number of times a given word occurs
in the document or the whole collection (i.e. the whole WWW).
However, the whole collection is not available to the fish-search,
so neither are these numbers.
In case a regular expression is given instead of a set of words
the fish-search simply counts the number of matches, relative to the
size of the document.
The fish-search offers the agrep syntax and library which can
tolerate errors in the given pattern, possibly meaning that it can find
matches even where the document contains typos.