DeepGraphLearning/KnowledgeGraphEmbedding

Why start = 4

passermyh opened this issue · 0 comments

Hello, I'm a student who try to learn from this paper and code, and I met a small question confused me.

def count_frequency(triples, start=4):
'''
Get frequency of a partial triple like (head, relation) or (relation, tail)
The frequency will be used for subsampling like word2vec
'''
count = {}
for head, relation, tail in triples:
if (head, relation) not in count:
count[(head, relation)] = start

why start = 4 instead of start = 1?