Where can I get more information about how the SCORE is computed?

Environment:    Oracle 12.1.0.2 EE on Exadata

I am confused about how Oracle Text computes the SCORE it returns after a simple query using two (2) domain indexes and a simple CONTAINS clause.

select score(1), etc. from <table> where CONTAINS(<document_column>,'keyword keyword keyword', 1) > 0;

In my particular case I do a UNION of a search of the column of document names with a search of the column of the documents themselves (.docs, .xlsx, .pdf).

The scores that are returned are a bit confusing to me.

The score of a document with the EXACT title of the above 3 words returns a lower score than other documents that contain the 3 words as a phrase and in many cases several times.