根据同义词在MySQL中进行高级搜索

Question

我有三个表，下面是相关的详细信息

1. Primary words

+----+---------+
| id | word    |
+----+---------+
| 1  | Machine |
+----+---------+
| 2  | phone   |
+----+---------+

2. Alternative words    

+----+------------+-----------+
| id | primary_id | word      |
+----+------------+-----------+
| 1  | 1          | system    |
+----+------------+-----------+
| 2  | 1          | feature   |
+----+------------+-----------+
| 3  | 2          | telephone |
+----+------------+-----------+

3. product table

+----+------------------+
| id | name             |
+----+------------------+
| 1  | mobile system    |
+----+------------------+
| 2  | computer machine |
+----+------------------+
| 3  | wired telephone  |
+----+------------------+

现在的转折是，每当用户在产品表中使用“机器”搜索时，会显示表名称为“机器”或“系统”或“功能”的产品结果，如果使用“系统”或“功能”搜索，那么也显示“机器”或“系统”或“功能”的结果。反之亦然。

您能建议我如何解决这个问题吗？

Answer 1

如果我理解正确你在问什么......

SQL

SELECT * 
  FROM Product P
 WHERE EXISTS(SELECT fw.Word  --(6) select all products that exist in this query
                FROM (SELECT pw.Word --(1) select all Primary words matching input
                        FROM PrimaryWords pw
                       WHERE pw.Word = 'machine'
                      UNION --(3) union the results from both selects
                      SELECT aw.Word --(2) select all Alternative words that match input or have its primary matching it
                        FROM PrimaryWords pw INNER JOIN AlternativeWords aw
                          ON pw.Id = aw.PrimaryId
                       WHERE pw.Word = 'machine'
                          OR aw.Word = 'machine') as fw --(4) alias the result
                WHERE p.Name LIKE '%' || fw.Word || '%'); -- (5) filter products that match the valid words

您可以读取按 () 中编号排序的 cmets。

Answer 2

首先你需要将这两个表数据合并为一个，然后给出条件。

Select word from(
SELECT id,word FROM PrimaryWords
Union all
Select primaryid,word from AlternativeWords a
) where id in ( select id from primarywords where word='yoursearchketword'
Union
Select primaryid from AlternativeWords where word='yoursearchketword')

根据您的产品表更新答案。

现在你需要交叉连接产品表，因为它们之间没有任何关系。

还有一件事是您必须在此处使用like 运算符将您想要的结果与产品表的名称列进行比较。在这里，我给出了如何实现的小想法，但您可以轻松地改进它。

Select a.word,b.name from
  (Select word from(
    SELECT id,word FROM PrimaryWords
    Union all
    Select primaryid,word from AlternativeWords a
    ) where id in ( select id from primarywords where word='yoursearchketword'
    Union
    Select primaryid from AlternativeWords where word='yoursearchketword')) a, product b
Where a.word LIKE CONCAT('%',name, '%');

Answer 3

我建议把它放在一个这样的表中：

id | primary_id | word
1 | 1 | machine
2 | 2 | phone
3 | 1 | system
4 | 1 | feature
5 | 2 | telephone

查询是这样的，我们以machine这个词为例：

Select word from MyTable where primary_id in 
             (select primary_id from MyTable where word = "machine")

输出：machine、system、feature

---

更新

如果您有更多列，例如为了显示这些词的相关程度，您可以再添加一个表来映射词之间的关系，如下所示：

id | w1_id | w2_id | relation
1 | 1 | 3 | 80
2 | 2 | 5 | 90
3 | 1 | 4 | 60
4 | 3 | 4 | 60

单词表将仅列出以下单词：

id| word
1 | machine
2 | phone
3 | system
4 | feature
5 | telephone

在this thread 中，您会发现一个很长的讨论，讨论您必须设计架构布局的方式以及如何检索数据的提示。