I imagine you need the data science to discern semantically relevant from irrelevant signals. How else do you “tell” your model what to look for? You could easi

I imagine you need the data science to discern semantically relevant from irrelevant signals. How else do you “tell” your model what to look for? You could easily train for an irrelevant but fitting model.