with q(val, rn) as (select '23/07/2019 17:15, ИВАНОВ Сергей Петрович', 1 from dual union all
select '23/07/2019 O7:15:41, Фернандес Перес Владимир Кастулович ',2 from dual union all
select '2/07/2019 о7:15, Николаев николай Николаевич',3 from dual union all
select '23/07/2019 17:10, Иванов Сергей',4 from dual union all
select '23/07/2019 17:19, Бендер Остап Сулейман-Берта - Мария Ибрагимович',5 from dual
)

select q.*
     , replace(
               listagg(
                       regexp_substr(val
                                     ,decode(
                                             level,1,'([[:alpha:]][[:alpha:]]*\s)','\s([[:alpha:]])[[:alpha:]*(\s|$)]'
                                             )
                                     ,1,level,null,1
                                     )
                       ,'.'
                       ) WITHIN GROUP (order by level)||'.'
               ,' .',' '
               )
from q
connect by regexp_count(val,'[[:alpha:]][[:alpha:]](\s|$)') >= level and prior rn = rn
and prior dbms_random.value is not null
group by rn, val
order by rn

decode(level,1,'([[:alpha:]][[:alpha:]].*?\s)','\s([[:alpha:]])[[:alpha:]*?(\s|$)]')