The ranking algorithm is somewhat complex, plus every aspect can be
tailored. see: http://www.thunderstone.com/texisman/node59.html
a discussion of the Texis 'likep' ranking search.
The following are considered:
Matching term cluster proximity
Matching term cluster ordering
Matches in document vs freq in corpus
Freq of term/phrase in corpus
Cluster distances from begin of indexed field(s)
The text index in Webinator is created against the Title\Meta\Body
combination of fields. This means that with the default 'lead-biasing'
factor the Title has the most importance, then Meta data, then the page
content. What this means is that if all other factors about a result
ranking are identical then Webinator will deem an answer in the
Title with a higher weight than one in the body of the html.