Eva: node attribute weighting/importance & support for continuous attributes

[eqo295]

Hello,
I am currently working on clustering a web network based on visitor information.

From my understanding, the Eva algorithm enables community detection while incorporating multiple node attributes (in this case, each node represents a website).

I have a couple of related questions:

1. Node attribute weighting / importance

Is there a way to:

• explicitly control or adjust the relative weights of different node attributes in Eva, or
• measure and compare how much each attribute influences the resulting communities (e.g., through metrics, scores, or internal variables)?

2. Support for continuous node attributesAs far as I know, Eva is designed to work with nominal metadata (single or multiple).

• Currently, I am discretizing continuous-valued attributes into categorical bins as a workaround.
• Does CDlib provide any community detection algorithms that can directly handle continuous node attributes, or is discretization the recommended approach?

Thank you for your time and for maintaining CDlib.

[eqo295]

Additional questions / updates:

1. My network is a weighted, undirected graph.

   • Does Eva support edge weights?

2. I also tried running the example code for iLouvain based on the current documentation:
   https://cdlib.readthedocs.io/en/latest/reference/generated/cdlib.algorithms.ilouvain.html

   However, I encountered execution errors when running the example.

   • Is iLouvain currently maintained and supported?
   • Are there any known issues or dependency/version constraints?