Question

我对索引不是很熟悉。我正在对记录应用索引，因为它花了很多时间在表上查找结果。查询工作正常

SELECT words.wordid, synsets.synsetid, pos, definition
FROM words, senses, synsets
WHERE words.wordid = senses.wordid
AND senses.synsetid = synsets.synsetid
AND lemma='apple'


| wordid | lemma               |                       
+--------+--------------------+-
|  1     | Apply               |
|  2     | PRIMARY             |
|  3     | DEPENDENT           |
+----+--------------------+-----

其他表格

wordid  casedwordid  synsetid  senseid  sensenum    laxid   tagcount
1         19        108659519   1       1       0   0
2       1-hitter    108970180   2       1       0   0
3        NULL       104510146   5       1       0   0

Synset的第3表

感官表

synsetid   pos     lexdomainid  definition
108659519   n       3   that which is perceived or known or inferred to have its
108970180   n       3   an entity that has physical existance   
104510146   n       3   a general concept formed by extracting common features

sql查询工作正常，但现在我必须应用索引来获得比简单查询更快的结果。

如何在这三个表上应用索引

Answer 1

如果你只运行SELECT，那么

SELECT words.wordid, synsets.synsetid, pos, definition
    FROM words, senses, synsets
    WHERE words.wordid = senses.wordid
    AND senses.synsetid = synsets.synsetid
    AND lemma='apple'

您需要这些索引（请参阅手册中的更多内容：http://dev.mysql.com/doc/refman/5.7/en/create-index.html）：

-- create simple b tree indexes for int fields (improve join performance)
CREATE INDEX words_wordid ON words (wordid);
CREATE INDEX senses_wordid ON senses (wordid);
CREATE INDEX senses_synsetid ON senses (synsetid);
CREATE INDEX synsets_synsetid ON synsets (synsetid);

-- we create an index using the first 20 characters of the lemma
CREATE INDEX words_lemma ON synsets (lemma(20));

如果要运行另一个选择查询，则应添加更多索引以提高选择性能：）

Answer 2

首先，让我们使用JOIN..ON语法：

SELECT  words.wordid, synsets.synsetid, pos, definition
    FROM  words
    JOIN  senses ON words.wordid = senses.wordid
    JOIN  synsets ON senses.synsetid = synsets.synsetid
    WHERE  words.lemma='apple'

现在，我们可以看到WHERE子句中唯一有用的是words.lemma。所以，让我们来words：

INDEX(lemma)

在查看了苹果words后，下一个表格为senses。现在该表需要：

INDEX(wordid)

最后要有效地进入synsets，请

INDEX(synsetid)

根据列名称，我猜测 synsetid已经是PRIMARY KEY。请注意，PRIMARY KEY是一个UNIQUE密钥，INDEX。所以，如果是PK，请不要添加我建议的INDEX。

Cookbook on indexing

如何使用Join在MySql中应用索引

感官表

2 个答案: