Question

I am making a synonyms list that I will store it in database and retrieve it before doing full text search.

When users enters like: word1

I need to lookup for this word in my synonyms table. So if the word is found, I would SELECT all the synonyms of this word and use it in the fulltext search on the next query where I contruct the query like

MATCH (columnname) AGAINST ((word1a word1b word1c) IN BOOLEAN MODE)

So how do I store the synonyms in a table? I found 2 choices:

  1. using key and word columns like

    val  keyword
    -------------
    1    word1a
    1    word1b
    1    word1c
    2    word2a
    2    word2b
    3    word3a
    etc.
    

So then I can find exact match of the entered word in one query and find it's ID. In the next select I get all the words with that ID and somehow concate them using a recordset loop in server side langauge. I can then construct the real search on the main table that I need to look for the words.

  1. using only word columns like

    word1a|word1b|word1c
    word2a|word2b|word2c
    word3a
    

Now I so the SELECT for my word if it is inside any record, if it is, extract all the record and explode it at | and I have my words again that I can use.

This second approach lookes easier to maintain for the one who would make this database of synonyms, but I see 2 problems:

a) How do I find in mysql if a word is inside the string? I can not LIKE 'word1a' it because synonims can be very alike in a way word1a could be strowberry and strowberries could be birds and word 2a could be berry. Obviously I need exact match, so how could a LIKE statement exact match inside a string?

b) I see a speed problem, using LIKE would I guess take more mysql take than "=" using the first approach where I exact match a word. On the other hand in the first option I need 2 statements, one to get the ID of the word and second to get all the words with this ID.

How would you solve this problem, more of a dilemma which approach to take? Is there a third way I don't see that is easy for admin to add/edit synonyms and in the same time fast and optimal? Ok I know there is no best way usually ;-)

UPDATE: The solution to use two tables one for master word and second for the synonym words will not work in my case. Because I don't have a MASTER word that user types in search field. He can type any of the synonyms in the field, so I am still wondering how to set this tables as I don't have master words that I would have ID's in one table and synonims with ID of the master in second table. There is no master word.

Was it helpful?

Solution

Don't use a (one) string to store different entries.

In other words: Build a word table (word_ID,word) and a synonym table (word_ID,synonym_ID) then add the word to the word table and one entry per synonym to the synonyms table.

UPDATE (added 3rd synonym)

Your word table must contain every word (ALL), your synonym table only holds pointers to synonyms (not a single word!) ..

If you had three words: A, B and C, that are synonyms, your DB would be

WORD_TABLE            SYNONYM_TABLE
ID | WORD             W_ID | S_ID
---+-----             -----+-------
1  | A                  1  |  2
2  | B                  2  |  1
3  | C                  1  |  3
                        3  |  1
                        2  |  3
                        3  |  2  

Don't be afraid of the many entries in the SYNONYM_TABLE, they will be managed by the computer and are needed to reflect the existing relations between the words.



2nd approach

You might also be tempted (I don't think you should!) to go with one table that has separate fields for word and a list of synonyms (or IDs) (word_id,word,synonym_list). Beware that that is contrary to the way a relational DB works (one field, one fact).

OTHER TIPS

I think 3 columns and only one table is better WORD_TABLE

ID | WORD | GroupID
---+----------------
1  |  A   |   1
2  |  B   |   1
3  |  C   |   1

Another approach is to store meaning (this does not use master words, but a meaning table that groups instead)

would be to store the words in a words table without synonyms and with only text, like this:

Many words, one meaning

meaning_table

meaning_id
---
1
2
3

And store the words in another table, for example if A, B and C were all synonyms of 1 meaning

word_table

word_id | meaning_id | word
--------+------------+------
1       |  1         |   A
2       |  1         |   B
3       |  1         |   C

Even though it looks a lot like what Hasan Amin Sarand suggests, it has the key difference that you don't select from the WORD_TABLE but instead select from the MEANING_TABLE, this is much better and I learned that the hard way.

This way you store the meaning in one table and as many words for that meaning as you like in another.

Although it assumes that you have 1 meaning per word.

Many words, many meanings

if you want to store words with multiple meanings then you need another table for the many to many relationship and the whole thing becomes:

meaning_table
-------------
meaning_id
-------------
1
2
3

word_meaning_table
--------------------
word_id | meaning_id
--------+-----------
1       |  1         
2       |  1        
3       |  1         

word_table
--------------
word_id | word
--------+-----
1       |   A
2       |   B
3       |   C

Now you can have as many words with as many meanings as you want, where any word can mean anything you want and any meaning can have many words.

If you want to select a word and it's synonyms then you do

SELECT 
meaning_id,word_id,word
FROM meaning_table 
INNER JOIN word_meaning_table USING (meaning_id)
INNER JOIN word_table USING (meaning_id)
WHERE meaning_id=1

You can also then store meaning that does not have a word yet or that you don't know the word of.

If you don't know what meaning it belongs to then you can just insert a new meaning for every new word and fix the meaning_id in the word_table later.

You can then even store and select the words that are the same but mean different things

SELECT 
meaning_id,word_id,word
FROM meaning_table 
INNER JOIN word_meaning_table USING (meaning_id)
INNER JOIN word_table USING (meaning_id)
WHERE word_id=1
Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top