Awesome

AutoAlign: Fully Automatic and Effective Knowledge Graph Alignment enabled by Large Language Models

This is our implementation for the paper:

AutoAlign: Fully Automatic and Effective Knowledge Graph Alignment enabled by Large Language Models

Description

The task of entity alignment between knowledge graphs (KGs) aims to identify every pair of entities from two different KGs that represent the same entity. Many machine learning-based methods have been proposed for this task. However, to our best knowledge, existing methods all require manually crafted seed alignments, which are expensive to obtain. In this paper, we propose the first fully automatic alignment method named AutoAlign, which does not require any manually crafted seed alignments. Specifically, for predicate embeddings, AutoAlign constructs a predicate-proximity-graph with the help of large language models to automatically capture the similarity between predicates across two KGs. For entity embeddings, AutoAlign first computes the entity embeddings of each KG independently using TransE, and then shifts the two KGs’ entity embeddings into the same vector space by computing the similarity between entities based on their attributes. Thus, both predicate alignment and entity alignment can be done without manually crafted seed alignments. AutoAlign is not only fully automatic, but also highly effective. Experiments using real-world KGs show that AutoAlign improves the performance of entity alignment significantly compared to state-of-the-art methods.

What are in this Repository

This repository contains the following main contents:

/
├── code/                         
|   |   ├── AutoAlign.py      --> (The main code of AutoAlign)
├── data/                   
|   ├── DW-NB/                --> (The DW-NB dataset)
|   ├── DY-NB/                --> (The DY-NB dataset)
|   ├── for_prox-graph/       --> (Necessary code for generating proximity graphs)
├── img/                      --> (The images for README (not used for the code))   

Run our code

Python >= 3.6.13

Need to set up GPU environment.

Download the corresponding versions of CUDA Toolkit and cuDNN.

CUDA version: 11.7

1. Install cudatoolkit and cudnn:

conda install cudatoolkit==9.0 cudnn==7.1.2

2. Install the required packages:

pip install -r requirements.txt

3. Run the code

python AutoAlign.py

Use your own datasets

To use your own datasets:

1. Prepare the KG files (each KG in one file) in Turtle（Terse RDF Triple Language）format (as the .ttl files in ./data/DY-NB/ folder).
2. Follow the instructions in ./data/for_prox_graph/ to generate the proximity graph.
3. Run on your own datasets in AutoAlign.py

Cite our paper

Please credit our work by citing the following paper:

@article{zhang2023autoalign,
  title={AutoAlign: Fully Automatic and Effective Knowledge Graph Alignment enabled by Large Language Models},
  author={Zhang, Rui and Su, Yixin and Trisedya, Bayu Distiawan and Zhao, Xiaoyan and Yang, Min and Cheng, Hong and Qi, Jianzhong},
  journal={arXiv preprint arXiv:2307.11772},
  year={2023}
}