search

QuanLab / node2vec-spark

Implement node2vec algorithm using Spark 2 from: http://snap.stanford.edu/node2vec/
11 stars 5 forks source link

Why is the value of each dimension of the embedding that I get so large? #1

Closed WangFeng666 closed 5 years ago

WangFeng666 commented 5 years ago

Hi, I'm trying to run node2vec using the Spark implementation on a large graph (~0.6M nodes, ~20M edges. Why is the value of each dimension of the embedding that I get so large?

企业微信截图_20191119171612

so please tell me why, thank you ! @QuanLab

QuanLab commented 5 years ago

Can you provide more detail about the problem or show a part of input? it would be helpful for me to solve this issue.

WangFeng666 commented 5 years ago

I can run, but embedding is not well. I wonder if it's because each dimension is too large. a part of input 1574229801(1)

WangFeng666 commented 5 years ago

The first column is id of users, and the second column is index for words(-1~-7000)

QuanLab commented 5 years ago

Read the document again and make sure that you are fully understand node2vec algorithm and what input you need. This issue will be closed.