Simple Interactive Evaluator

[JVonderwell]

We'd like a simple way for users to self-evaluate holoclean's performance in cases where we don't have a ground truth. This will amount to essentially just listing each change (or a subset of changes) to the user and asking them "did HC do the right thing?" and calculating simple precision metrics based off of that.

Benefit: Providing users and ourselves a basic understanding of HC's performance on a given dataset.

Cost: Maybe a couple hours of coding to structure and test.

[JVonderwell]

TODO: Week of 4/23