SQL比较类似的列

时间:2013-11-19 19:30:38

标签: sql case string-comparison

我有两列可能包含类似数据,我正试图弄清楚如何比较两者的相似之处。

TABLE:
PatientID | InsurnaceCompany1 | InsuranceCompany2
111111111 | Medicaid          | Medicaid - Part D
222222222 | Medicaid - ABC Co | Medicaid
333333333 | Medicaid - ABC Co | Medicaid - Part D
444444444 | MVP Insurance     | MVP
555555555 | Medicare          | Medicaid
666666666 | Medicaid          | MVP
777777777 | Private - MVP     | MVP Option

问题是, 如何比较两列以了解它们是否相似。

结果应该是这样的:

PatientID | InsurnaceCompany1 | InsuranceCompany2 | Result
111111111 | Medicaid          | Medicaid - Part D | TRUE
222222222 | Medicaid - ABC Co | Medicaid          | TRUE
333333333 | Medicaid - ABC Co | Medicaid - Part D | TRUE
444444444 | MVP Insurance     | MVP               | TRUE
555555555 | Medicare          | Medicaid          | FALSE
666666666 | Medicaid          | MVP               | FALSE
777777777 | Private - MVP     | MVP Option        | TRUE

我可以使用case语句检查以下几个选项,如:

case
    when InsuranceCompany1 like '%' + InsuranceCompany2 + '%' then 'TRUE'
    when InsuranceCompany2 like '%' + InsuranceCompany1 + '%' then 'TRUE'
    else 'FALSE'
end AS Result

但上述PatientID 333333333和777777777的测试失败。

为了澄清,我将如下定义“类似”:如果它们各自包含相同的单词,则两个术语相似。即使这个词是通用的(例如“保险”),这也适用。所以“保险公司B”类似于“保险公司A”,原因有两个:“保险”和“公司”。

2 个答案:

答案 0 :(得分:0)

我认为你的问题中暗含了许多特定于领域的知识(究竟是什么构成了“相似”的名字),而且你无法通过一般目的的比较来解决这个问题。您最好的选择可能是在您的查询中枚举您的特定要求,这样您就可以准确控制您所获得的内容:

select
  *
from
  patients
where
  InsuranceCompany1 like concat('%', InsuranceCompany2, '%') or 
  InsuranceCompany2 like concat('%', InsuranceCompany1, '%') or
  (InsuranceCompany1 like '%medicaid%' and InsuranceCompany2 like '%medicaid%') or
  (InsuranceCompany1 like '%medicare%' and InsuranceCompany2 like '%medicare%') or
  (InsuranceCompany1 like '%mvp%' and InsuranceCompany2 like '%mvp%') or
  ...etc...

http://sqlfiddle.com/#!2/9de7b/13

答案 1 :(得分:-2)

修改

DECLARE @t TABLE (PatientID INT, InsurnaceCompany1 NVARCHAR(50), InsuranceCompany2 NVARCHAR(50))

INSERT INTO @t 
VALUES 
(111111111,'Medicaid'          ,'Medicaid - Part D'),
(222222222,'Medicaid - ABC Co' ,'Medicaid'),
(333333333,'Medicaid - ABC Co'  ,'Medicaid - Part D'),
(444444444,'MVP Insurance'      ,'MVP'),
(555555555,'Medicare'           ,'Medicaid'),
(666666666,'Medicaid'           ,'MVP'),
(777777777,'Private - MVP'      ,'MVP Option')


SELECT PatientID, InsurnaceCompany1, InsuranceCompany2 ,
    CASE WHEN LEFT(InsurnaceCompany1,CASE WHEN CHARINDEX(' ', InsurnaceCompany1) = 0 
                    THEN LEN(InsurnaceCompany1) ELSE CHARINDEX(' ', InsurnaceCompany1) END ) 
            = LEFT(InsuranceCompany2,CASE WHEN CHARINDEX(' ', InsuranceCompany2) = 0 
                THEN LEN(InsuranceCompany2) ELSE CHARINDEX(' ', InsuranceCompany2) END ) 
                THEN 'TRUE' ELSE 'FALSE' END    AS RESULT 
FROM @t

结果集

PatientID   InsurnaceCompany1   InsuranceCompany2   RESULT
111111111   Medicaid            Medicaid - Part D   TRUE
222222222   Medicaid - ABC Co   Medicaid            TRUE
333333333   Medicaid - ABC Co   Medicaid - Part D   TRUE
444444444   MVP Insurance       MVP                 TRUE
555555555   Medicare            Medicaid            FALSE
666666666   Medicaid            MVP                 FALSE
777777777   Private - MVP       MVP Option          FALSE