Ontology-aware embeddings for organismal trait descriptions

Trait descriptions characterize how an animal looks, behaves or interacts. These descriptions are often represented as text and then manually mapped within an ontology for downstream analysis. Nonetheless, the cost of this manual mapping is not scalable.

I am working on a transformer model that embeds textual trait descriptions in a latent space that captures the notion of distance within an ontology.

I am excited to explore how this could enhance trait characteristics inferred from images. For example, by associating it with the ontological representation of the closest embedding.

No files, tests, or entry points are named. Start by clarifying the intended transformer model, ontology source, and relationship to image-inferred traits with the maintainers. Done would require an agreed implementation scope and validation showing that textual embeddings capture ontology distance and support trait inference from images.

Written by the indexing model from the issue text.