This collection of scripts accompanies "Inductive Entity Typing Alignment" approach designed by Giuseppe Rizzo, Marieke van Erp, Raphaël Troncy.
NERD-inductive is an approach to automatically align entity type taxonomies through the use of annotated data, reasoning from the fact that the implicit knowledge contained in the annotated data can be made explicit by a machine learning approach in order to obtain dynamic mappings.
This repository contains the following directories and files:
- data: directory containing a data sample and the logic for traversing OWL ontologis
- plots: directory containing precision, recall and F-measure plots for the learnt alignments as well as the R-code to generate the plots
- experiments: directory containing the code that was used to run the experiments
- mappings: directory containing the obtained mappings between the NERD extractor taxonomies and the two benchmark datasets
- LICENSE: information about the license under which this code is released