Other evaluation metrics

[lindsey98]

Hi, I like your paper. As for the evaluation metrics, I can see your are using recall@k to track the performance. May I ask have you tried to implement other metrics like NMI and MAP@R? Thanks!

[pierre-jacob]

Hi,

Thanks for your interest! I did implement a few more metrics such as mAP@R (but not NMI), although the code is not compatible with the version from this repository for two main reasons:

• I updated my personal repository to Tensorflow 2.2+, so I don’t have a code compatible with this one,
• I had to rethink how metrics were implemented (especially how to handle custom distances and how to evaluate metrics for larger datasets), hence, the whole computation process is not compatible with this one.

If you're willing to implement it though, I can share some code.