Learning from score

Systematizing Confidence in Open Research and Evidence (SCORE) aims to provide confidence scores to social science results so that they can be more reliably useful (specifically to DoD. It is a very large effort with an incredible amount of work put into generating reliable metrics and measures for papers.

The SCORE Phase 2 Claim Extraction Coding Manual provides a detailed discussion of some of the most related aspects of the project.

There is also a lengthy write-up from someone who took part in the coding for SCORE.

A few immediate questions I hope to answer:

1. has SCORE learnt any meta methods that will help us be better (faster or more accurate) at creating new columns?
2. can we easily use the columns they have created? (either in a form of finished data or as something we might encode when adding papers to Atlas?