Search Relevance %

Hi,

In our defined KM Search, the results are sorted by relevance. But the relevance % doesn't go with the matches displayed in the document abstract. A high relevance % has 2 matches vs a lower relevance has 4 matches.

Does anyone know how the following:

1) How is this relevance % is calculated?

2) Is it solely related to the indexing done by TREX?

Any help would be much appreciated. PD We have Trex 6.0.

Regards