如何在Oracle中使用UTL_匹配显示不同表中两列之间的差异？

我想比较Oracle DB中的两列，如果字符串不同，我想显示其他列中的差异。我知道我错过了什么。 例如：

SELECT A.CD_KEY01,
       A.TEXT_01,
       B.TEXT_02,
       UTL_MATCH.edit_distance_similarity(A.TEXT_01, B.TEXT_02) AS distance_similarity
FROM TB_TABLE_01 A 
JOIN TB_TABLE_02 B
ON A.CD_KEY01 = B.CD_KEY02

我得到的示例输出：

CD_KEY01  |  TEXT_01              |  TEXT_02               | DISTANCE_SIMILARITY 
   111    |  Superman is good     |  Superman is good      | 100
   222    |  Superman is bad      |  Superman is bad       | 100
   333    |  Superman is handsome |  Hulk is ugly          | 33
   444    |  Superman is awful    |  Batman is awful       | 90

我需要的输出示例：

CD_KEY01  |  TEXT_01              |  TEXT_02               | DISTANCE_SIMILARITY | DIFF_01 | DIFF_02 
   111    |  Superman is good     |  Superman is good      | 100                 | NULL    | NULL
   222    |  Superman is bad      |  Superman is bad       | 100                 | NULL    | NULL
   333    |  Superman is handsome |  Hulk is ugly          | 33                  | Hulk    | ugly
   444    |  Superman is awful    |  Batman is awful       | 90                  | Batman  | NULL

---

我怀疑有没有简单的方法。一种方法是以某种递归方式将字符串拆分为单词，逐字比较，并以透视结果为准：

with 
  a(key, text_a, word, rn) as (
    select cd_key01, text_01, regexp_substr(text_01, '(\w+)', 1, 1), 1 
      from table_01 
    union all
    select key, text_a, regexp_substr(text_a, '(\w+)', 1, rn + 1), rn + 1 
      from a 
      where regexp_substr(text_a, '(\w+)', 1, rn + 1) is not null),
  b(key, text_b, word, rn) as (
    select cd_key02, text_02, regexp_substr(text_02, '(\w+)', 1, 1), 1 
      from table_02 
    union all
    select key, text_b, regexp_substr(text_b, '(\w+)', 1, rn + 1), rn + 1 
      from b where regexp_substr(text_b, '(\w+)', 1, rn + 1) is not null)
select *
  from (
    select key, rn, text_a, text_b, 
           case when a.word <> b.word then b.word end word 
      from a full join b using (key, rn))
  pivot (max(word) for rn in (1 w1, 2 w2, 3 w3)) order by key

此查询显示前三个单词的比较，is也进行比较。如果字符串可能有不同的字数，则在正确处理空部分时，必须小心并修改大小写