Question

I have a table called words, consisting of three columns word(VARCHAR(16)), doc_id(INT), weight(DOUBLE).

Here is what I need to do, I have two queries:

SELECT doc_id, weight FROM words WHERE word = 'bla';

doc_id    weight
------    ------
1         0.14
2         0.61
3         0.32

and

SELECT doc_id, weight FROM words WHERE word = 'blabla';

doc_id    weight
------    ------
2         0.19
3         0.45
4         0.14

I need to get the intersection of the two on doc_id and select the lower weight value as the weight, i.e. I want the results to be:

doc_id    weight
------    ------
2         0.19
3         0.32

Is there a way to do that in a single query? Doing it in the program makes it damn slow!

I also need to get their UNION and select the higher weight value, i.e. I want the results to be:

doc_id    weight
------    ------
1         0.14
2         0.61
3         0.45
4         0.14

Keep in mind that the column word and doc_id are not unique, so one word can be assigned to many docs.

Answer 1

For the intersect part it sounds like you want “the lowest weight for all doc_id where the doc_id has one row for the word ‘bla’ AND one row for the word ‘blabla'”. That can be found by

(untested)
select w1.doc_id, least(min(w1.weight), min(w2.weight)) as minweight 
  from words w1, words w2
 where w1.doc_id = w2.doc_id
   and w1.word = 'bla'
   and w2.word = 'blabla'
 group by w1.doc_id;

For the union part what you want is “the highest weight for all doc_id where the doc_id has one row for the word ‘bla’ OR one row for the word ‘blabla'”. That can be found by

(untested)
select doc_id, max(weight) as maxWeight
  from words
 where word in ('bla', 'blabla')
 group by doc_id;