When a document satisfies a text search expression a score is computed to reflect the closeness of the match. This is a positive integer, with a higher value meaning a closer match. The scores are only comparable between results of the same query against the same database. Moreover, the scoring rules are different for different versions of Virtuoso server, due to the progress in information retrieval theory. Thus the only really safe thing to do with scores is to sort the list of hits by descending score to produce more relevant hits first. In addition, the CONTAINS predicate allows specifying a lower limit to the score of produced hits so a smart application can reduce a number of retrieved documents by filtering out the less relevant documents.

While the real scoring rules of the server are too sophisticated to be listed here, the basic concept is simple: hits for restrictive terms are most important than hits for generic terms, frequent hits in same document are more important than occasional, hits that are probably concentrated in same sentence are more important than hits here and there in a long text:

The XCONTAINS predicate can also return scores. These scores are somewhat similar to scores made by CONTAINS but rules for them are too complicated to be explained here.