How is missing attribute handling implemented in the RAKGE model?

[jepril]

Hi there,

First of all, thank you for sharing this fascinating research and the code for the RAKGE model. I have been studying your paper "Exploiting Relation-aware Attribute Representation Learning in Knowledge Graph Embedding for Numerical Reasoning" and examining the code you provided on GitHub.

In the paper, you mentioned handling missing attribute values by employing a learnable missing value embedding. However, after reviewing the code, I couldn't find the implementation for dealing with missing numerical literals. Could you kindly point out where the handling of missing values is implemented? If it's not currently included in the code, could you offer some guidance on how I might implement it as described in the paper?

Thank you very much for your time and assistance!

Best regards,
jepril