The hypernym-hyponym and meronym-holonym relations have some discrepancies in the sense that if A is a hypernym of B, then B should b a hyponym of A and similarly with meronym-holonym but it is not the case. These discrepancies are also there in the database (the dropbox link in the constanst.py file)
a basic example:
the total number of direct hyponym relations is: 30884
the total number of direct hypernym relations is: 3530
another example:
the total number of meronym (component object) relations is: 718
the total number of holonym (component object) relations is: 714
The two numbers should be the same in both the case (as is the case in English wordnet provided by nltk)
code for finding hypernyms:
num_hypernym=0
for v in iwn.all_synsets.() :
num_hypernym += len(iwn.synset_relation( v , pyiwn.SynsetRelations.HYPONYMY))
Not sure if this is same in the original hindi/indo wordnet database or it is specific to pyiwn.
The hypernym-hyponym and meronym-holonym relations have some discrepancies in the sense that if A is a hypernym of B, then B should b a hyponym of A and similarly with meronym-holonym but it is not the case. These discrepancies are also there in the database (the dropbox link in the constanst.py file)
a basic example: the total number of direct hyponym relations is: 30884 the total number of direct hypernym relations is: 3530
another example: the total number of meronym (component object) relations is: 718 the total number of holonym (component object) relations is: 714
The two numbers should be the same in both the case (as is the case in English wordnet provided by nltk)
code for finding hypernyms: num_hypernym=0 for v in iwn.all_synsets.() : num_hypernym += len(iwn.synset_relation( v , pyiwn.SynsetRelations.HYPONYMY))
Not sure if this is same in the original hindi/indo wordnet database or it is specific to pyiwn.