Design API to enable semi-automatic classification

[nathanshammah]

We could integrate with an existing tool (that behind uses a language model engine) to enable semi-automatic classification of submissions. For example, upon linking of a submission, an API could scrape the abstract to help classify the paper based on buckets provided by Metriq (such as suggesting the most relevant top-level Task option among "Hardware", "Simulation", "Compiler", and "QEM/QEC"). More simply, it could also suggest what tags to add, in the modal of the submission.

I believe it's used for data annotation in the training process, but I wonder if Argilla would be a useful tool to plug in, https://argilla.io/