I have a use case where I have a word and I need to know the following things:
- Synonyms for the word (just the synonyms are sufficient)
- All senses of the word, where each sense contains - the synonyms matching that word in that sense, example sentences in that sense (if there), the part of speech for that sense.
Example - this query link. Screenshot for the word carry
:
For each 'sense', we have the part of speech (like V
), synonyms matching that sense, (like transport
in the first sense, pack
, take
in the second sense, etc), example sentences containing that word in that sense (This train is carrying nuclear waste
, carry the suitcase to the car
, etc in first sense, I always carry money
etc in the second sense, etc.).
How do I do this from a Wordnet MySQL database? I ran this query, it returns the list of meanings for the word:
SELECT a.lemma, c.definition FROM words a INNER JOIN senses b ON a.wordid = b.wordid INNER JOIN synsets c ON b.synsetid = c.synsetid WHERE a.lemma = 'carry';
How do I get the synonyms, example sentences, part of speech and synonyms specific to that sense for each sense? I queried the vframesentences
and vframesentencemaps
tables, saw example sentences with placeholders like %s
, and based on the wordid
column I tried to match them with the words
table, but got awfully wrong results.
Edit:
For the word carry
, if I run these queries, I get synonyms and sense meanings correctly:
1. select * from words where lemma='carry' //yield wordid as 21354
2. select * from senses where wordid=21354 //yield 41 sysnsetids, like 201062889
3. select * from synsets where synsetid=201062889 //yields the explanation "serve as a means for expressing something"
4. select * from senses where synsetid=20106288` /yields all matching synonyms for that sense as wordids, including "carry" - like 21354, 29630, 45011
5. select * from words where wordid=29630 //yields 'convey'
So all I now need is a way of finding the example sentence for the word carry
in each of the 41 senses. How do I do it?
You can get the sentences from the
samples
table. E.g:yields:
So you could extend your query as follows:
Note: The subselect with a
GROUP_CONCAT
returns the synonyms of each sense as a comma-separated list in a single row in order to cut down on the number of rows. You could consider returning these in a separate query (or as part of this query but with everything else duplicated) if preferred.UPDATE If you really need synonyms as rows in the results, the following will do it but I wouldn't recommend it: Synonyms and example sentences both pertain to a particular definition so the set of synonyms will be duplicated for each example sentence. E.g. if there are 4 example sentences and 5 synonyms for a particular definition, the results would have 4 x 5 = 20 rows just for that definition.