Thank you for your suggestions! All of these seem to be present already in our tokenization method, but I will double check later.
We avoided these settings on purpose, instead the tokenize process will look for some special characters (like hyphens, dashes, quotes etc..) and indexes keywords both with and without them. This was mostly implemented during the years via customer reports. You probably seen some of this as junk data in the database, but it is actually not.