SQL效率优化，O(n^2)时间复杂度，如何简化？

Question

insert into probclass (jobid, tag, probability) 
    (select s1.jobid, s1.tag, (select count(*) from sqlset as s2 
                             where s1.jobid = s2.jobid and s1.tag = s2.tag)
                            /
                            (select count(*) from sqlset as s3 
                             where s1.jobid = s3.jobid) 
     from sqlset as s1)

我是SQL的新手，但我正在尝试编写一个尽可能简单的查询。这是一个简单的查询，它计算具有相同'jobid‘的attr 'tag’的概率。我知道这是糟糕的编码，因为枚举数和分子数一次又一次地重复计算(即每次都是如此)。

有没有简单的O(n)查询？(简单地说，我的意思是不要使用很多子句，如WITH等。)

Answer 1

试着简单地说

     select count(*) as 'count1' into #temp1 from sqlset as s2 
     where s1.id = s2.id and s1.tag = s2.tag

和

     select count(*) as 'count2' into #temp2 from sqlset as s3 
     where s1.id = s3.id

如果这两者对于不同表是不同的计数值，那么

     insert into probclass (id, tag, probability) 
    (select s1.id, s1.tag, (select count1 from #temp1/select count2 from #te,p2))