Mistake when calculating Rel statistics

[williamLyh]

Hi!
I believe there's a mistake in stats.py line 23.
The eid is the object_id, which could correspond to more than one object synset, but here the first synset will be overwritten by the second synset (or the last one if there are more than 2). This leads a decrease in information in the returned entities dictionary, which means when calculating the relation, only one of the synset is considered.
For example, when object_id=162759, the objects are ['position.n', 'ocean.n'], but entities['162759'] is only 'position.n'
I think this could be a mistake and you might want to handle this more carefully.

Best,
Yinhong Liu