with a as
(
select '1234' key1,    '1' key2,    'Y' data from dual union all
select '1234' key1,    '2' key2,    'Y' data from dual union all
select '1234' key1,    '3' key2,    'Y' data from dual union all
select '4567' key1,    '0' key2,    'Y' data from dual
), b as
(
select '1234' key1,    '1' key2,    'Y' data from dual union all
select '1234' key1,    '3' key2,    'N' data from dual union all
select '8743' key1,    '0' key2,    'Y' data from dual
)
select  a.*,
        case when a_data = b_data then '전부동일'
             when cnt > 1 and a_key2 is not null and b_key2 is null then 'key2없음'
             when nvl(a_data,b_data) != nvl(b_data,a_data) then 'data값 틀림'
             when a_key1 is not null and b_key1 is null then 'a table only'
             when a_key1 is null and b_key1 is not null then 'b table only'
             else '기타'
        end  결과
from (
select  a.key1 as a_key1,
        a.key2 as a_key2,
        a.data as a_data,
        b.key1 as b_key1,
        b.key2 as b_key2,
        b.data as b_data,
        count(*) over (partition by nvl(a.key1,b.key1)) cnt    
from a full outer join b
     on  a.key1 = b.key1
     and a.key2 = b.key2
) a
order by nvl(a_key1,b_key1),nvl(a_key2,b_key2)