delete from my_table  a
where
   a.rowid <>
     (select max(rowid) from my_table b
      where b.key1 = a.key1
      and b.key2 = a.key2
      and b.key3 = a.key3); 

create table rebuild_my_table as
select distinct *
from my_table;

truncate table my_table;

insert into my_table select * from rebuild_my_table;

commit;

drop refresh_my_table;