prokolyvakis / deep-align

This repository contains our implementation of the ontology matching framework based on representation learning.

Biomedical Ontology Alignment: An Approach Based on Representation Learning

This repository contains our implementation of the ontology matching framework based on representation learning.

License

Apache License Version 2.0. For more information, please refer to the license.

Instructions for running:

• Prerequisites :

  • Python, Project Jupyter.
  • Python Libraries: NumPy, SciPy, scikit-learn, Theano, Lasagne, Keras, NLTK, pickle, PyYAML.

• Preprocessed Data:

  • All the data used can be found in data/ directory.
  • In each file, in the data/ directory, there are six different files. Specifically, there is:
    1. A .yaml file that stores the configuration parameters.
    2. A file named training_data contains the training data.
    3. A file name pretrained-wikipedia-pubmed-and-PMC-w2v contains the pre-trained word vectors.
    4. Two files named terms_of_ontology_1 and terms_of_ontology_2 contain the terms of the ontology 1 and ontology 2, respectively.
    5. A file named ground_truth_alignments contains the ground-truth alignments.

• Perform the different ontology matching tasks:

  • Launch the Jupyter Notebook App and execute the Notebook document Ontology_Matching.

Contact

• prodromos DOT kolyvakis AT epfl DOT ch