I’m working on a project that requires multi-label hierarchical classification, where each level of the hierarchy has multiple classes to predict. Specifically, I’m dealing with a scenario where the labels follow a hierarchical structure, and at each level of the hierarchy, there are multiple possible classes.

My project is aimed at classifying Instagram users based on their interests. The goal is to categorize users into hierarchical interest groups, where each user may belong to multiple categories at different levels of the hierarchy.

For instance, considering Cristiano Ronaldo as an example user, his primary interest might be in “Sports” and “family and friends” as top-level categories , with further refinement of the “sports” category into “Soccer” and “Handball” (hypothetically) as subcategories at the same level.

I’ve explored some existing methods for hierarchical classification, but they seem to focus on single-label classification or assume a single class at each level. Can anyone suggest an approach or point me towards resources for implementing multi-label hierarchical classification with multiple classes at each level?

Any insights, algorithms, or code examples would be greatly appreciated.
Thanks in advance for your help!

I’ve been experimenting with multi-label hierarchical classification for categorizing Instagram users based on their interests. However, I’ve encountered a limitation in my approach: it seems to only classify one class at each level of the hierarchy, which doesn’t fit my use case.