方案一：wm_concat函数
select username, id, wmsys.wm_concat(subject) as subject, wmsys.wm_concat(score) as score
from STUDENTSCORES
group by username, id

方案二：listagg函数
select username, id, LISTAGG(subject, ‘-‘) within group(order by subject) as subject, LISTAGG(score, ‘,‘) within group(order by score) as score
from STUDENTSCORES
group by username, id

方案三：常规sql
select username, id, translate(ltrim(subject, ‘/‘), ‘*/‘, ‘*,‘) as subject,translate(ltrim (score, ‘/‘), ‘*/‘, ‘*,‘) as score
from 
(select row_number() over (partition by username, id order by username, id, lvl desc) as rn, username, id, subject, score
from 
(select username, id, level lvl,                            sys_connect_by_path (subject, ‘/‘) as subject, sys_connect_by_path (score, ‘/‘) as score
from 
(select username, id, subject, score,                                       row_number() over (partition by username,id order by username, id) as num from STUDENTSCORES order by username, id)
connect by username = prior username and id = prior id and num - 1 = prior num))
where rn = 1;

注意：
方案一中默认分隔符为 ‘,’
方案二只适合11g之后的版本

Oracle 多行数据合并成一行