We describe a method for representing knowledge graphs that capture an intensional representation of the original extensional information. This representation is very compact, and it abstracts away from individual links, allowing us to find better path candidates, as shown by the results of link prediction using this information. The data in this archive consists of the abstract graphs built from several configurations of Freebase and NELL (used in the experiments described in Gardner et al. 2014), the abstract paths extracted from these graphs, the grounded paths and the negative sampling used in the experiments described in (Nastase and Kotnis, 2019). The motivation for this work was to find better patterns in knowledge graphs than those obtained using the PRA approach. This dataset contains:

 AbstractGraphs_Garder2014data.tar.gz: the abstract graphs AbstractPathsGroundedPaths.tar.gz: the abstract paths, grounded paths, train/test data and the corresponding negative samples (for each abstract graph)

For a brief description of the data in each archive, please see README_AbstractGraphs.txt.