This release corresponds to the Automatic Alignment Model 1.0.
This first version is used in our paper "Aligning Actions Across Recipe Graphs". In this paper, we present two automatic alignment models (base, extended) and a simple baseline (cosine similarity).

This first version was trained and tested on the ARA 1.0 version of the ARA corpus. The updated ARA 1.1 version and an explanation about the differences is available here.