Add MACE

[jcklie]

Is it possible that you add MACE ? It is often used in my field but there is only a Java implementation that is hard to integrate into Python projects.

[dustalov]

Thank you for reaching us! I have added MACE to our internal roadmap. Since this method is similar to the already present Dawid-Skene and to-be-added-soon GLAD methods, could you please provide us with an example showing a significant quality boost?

[jcklie]

Thank you! I think the performance between the two is pretty similar (see this evaluation), the main difference from what I see is that D&S provides a confusion matrix as a side result whereas MACE shows the spamming preference and the credibility of annotators.

[dustalov]

There is a single-coin variation of the Dawid-Skene model that uses reliability levels instead of confusion matrices. It is in our roadmap, too. We will consider adding MACE, but from what I know, it is not always enough to model spamming preference in a single parameter. In that evaluation paper, MACE was on par with Dawid-Skene (thank you for the reference).

[dustalov]

Added by @pilot7747. @jcklie please try it out!