search

benedekrozemberczki / ASNE

A sparsity aware and memory efficient implementation of "Attributed Social Network Embedding" (TKDE 2018).
GNU General Public License v3.0
82 stars 27 forks source link

predict function #2

Closed ARNABKUMARPAN closed 5 years ago

ARNABKUMARPAN commented 5 years ago

can you please provide the predict function?

benedekrozemberczki commented 5 years ago

?

On Thu, 4 Apr 2019 at 12:02, ARNABKUMARPAN notifications@github.com wrote:

can you please provide the predict function?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/benedekrozemberczki/ASNE/issues/2, or mute the thread https://github.com/notifications/unsubscribe-auth/AQk2Fx6fQXJxY4S66jLeBo2--9EjP1Y6ks5vddu5gaJpZM4cchTH .

ARNABKUMARPAN commented 5 years ago

Hello, How to use the embedding that you have generated for link prediction? Is there any implementation of that? I am new in node embeddings. I will be very happy if you kindly provide me the understanding along with code. Thanks and Regards Arnab

On Thu, Apr 4, 2019, 4:33 PM Benedek Rozemberczki notifications@github.com wrote:

?

On Thu, 4 Apr 2019 at 12:02, ARNABKUMARPAN notifications@github.com wrote:

can you please provide the predict function?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/benedekrozemberczki/ASNE/issues/2, or mute the thread < https://github.com/notifications/unsubscribe-auth/AQk2Fx6fQXJxY4S66jLeBo2--9EjP1Y6ks5vddu5gaJpZM4cchTH

.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/benedekrozemberczki/ASNE/issues/2#issuecomment-479852263, or mute the thread https://github.com/notifications/unsubscribe-auth/AYrXF5JaaXELrs_0SeIoraIwBbfaZp4Mks5vddwWgaJpZM4cchTH .

benedekrozemberczki commented 5 years ago
  1. You have to create a corrupted graph by removing edges, but maintaining the number of connected components. The removed edges are positive cases.
  2. You have to create an embedding of the corrupted graph.
  3. You have to create a new graph where there are additional fake edges. The fake edges are negative cases.
  4. Calculate for each edge in the augmented graph using the embedding of the attenuated graph the representation -- use binary operators described in Node2Vec. This is the work of (Grover and Leskovec , 2016).