New nodes in recommender system

Questions
#1

Hi everyone.

I am planning to build a recommender system following this guidelines: 6.3 Training GNN for Link Prediction with Neighborhood Sampling — DGL 0.6.1 documentation

It would be a recommender system with users and products as nodes. My question is, after the training, what is the procedure to follow when the system is already built but new users or products have to be added to the original training dataset?

Thanks!

#2

You will need to:

  1. Extract node features for new user and product nodes
  2. Add them to the original graph

If these new user and product nodes are all isolated nodes, you can still try applying your model, but the performance may not be good.

2 Likes
#3

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.