What is the purpose of the SPOC Verifier?

[cncoleman]

The Streamlit app we are designing as SPOC Verifier has the potential to serve multiple purposes:

1. If we have no SPOC training data at all (3 data points: Species-time-place), we can use the matched species entities in the papers, see them in context, match them to GDIF, etc. and, bring all that data together in one place to make it easier for experts to complete the three data points
2. If we have a pre-trained model (with data from COPIOUS?) but need to tune it to our data set, SPOC Verifier can be used to evaluate, correct, and verify to generate an improved training set.
3. Once we have our own SPOC trained model, we can't expect it to ever be perfect. We'll always need to verify the results before we are prepared to submit them. The SPOC verifier will allow us to see where there are results but low confidence, where there are remaining gaps, and where there are outliers.

[cncoleman]

In other words, the very particular nature of this data -- where every record must be accurate -- means that human oversight will always be required. It's not just for creating training data. Yet the required tools/features are very much the same up until the final export. The beauty is that the model can be shared and reused as can the verifier.