Entity Alignment between Knowledge Graphs

An open-source framework for aligning the same real-world entities in different knowledge graphs.

Content

1. KBA.ipynb
2. data/

Requirement

1. Python (>=2.7)
2. Numpy (>=1.13.3)
3. TensorFlow (>=1.4.1)
4. CUDA (>=9.0)
5. Matplotlib (>=2.0.0)
6. scikit-learn (>=0.18)

Predicate alignment

To obtain the best entity alignment performance, our model requires a predicate alignment pre-processing step as follows:

1. Install python-Levenshtein

    pip install python-Levenshtein
   

2. Compute the similarity of all possible predicate pairs between two KGs. To compute the similarity, you can use the following command:

    import difflib
    ...
    pred_similarity=difflib.SequenceMatcher("predicate_1", "predicate_2").ratio()
    ...
   

3. Filter the predicate similarity (pred_similarity) using a certain threshold. We use the threshold of 0.95.

4. Manually check the final result.

References

Bayu Distiawan Trisedya, Jianzhong Qi, Rui Zhang. [Entity Alignment between Knowledge Graphs Using Attribute Embeddings.] AAAI 2019.