Assessing ML methodologies to Extract Implicit relations from documents Challenge

The goal of the proposed challenge is to assess the current situation concerning Machine Learning (ML) algorithms for Information Extraction (IE) from documents, identifying future challenges and to foster additional research in the field. The aim is to:

• Define a methodology for fair comparison of ML algorithms for IE.
• Define a publicly available resource for evaluation that will exist and be used beyond the lifetime of the challenge; such framework will be ML oriented, not IE oriented as so far proposed in other similar evaluations.
• Perform actual tests of different algorithms in controlled situations so to understand what works and what does not and therefore identify new future challenges.