I have a frequent instances of segments being split where roman numerals are used before or after abbreviations, where the roman numerals are not recognised as such, in particular in relation to two commonly used citation forms in Austrian law.

The most frequent cases are:

BGBl. I (i.e. volume I of the Federal Law Gazette) - most commonly in the form BGBl. [I|II|III] xxx/yyyy
BGBl. II (i.e. volume II of the Federal Law Gazette)

I. - XXVII. GP. (i.e. 1st to 27th legislative period) - in some documents 1. - 27. GP. xxx

particularly where there is no non-breaking space between the roman numeral and BGBl. or GP.

How can I handle this issue using RegEx and segmentation rules?