WinDB3ll / OpenHHEA

An open-source highly heterogeneous entity alignment (HHEA) toolkit.

OpenHHEA : Toolkit for Entity Alignment

Code of OpenHHEA

Requirements

python >= 3.9
scipy
numpy
networkx
pillow
gensim
openai
transformers
torch

Main Interface

The main interface provided by OpenHHEA is included as follows:

• OpenHHEA.get_model_config: get the predifined configurations of model and data processing.
• OpenHHEA.KGDataLoader: the data loader for loading knowledge graph data.
• OpenHHEA.KGDataProcessor: the processor for process the knowledge graph data to get feature embeddings of name information, structure information, image information and so on.
• OpenHHEA.HHEATrainer: the trainer for training embedding-based entity alignment model. Use train function to start training.
• OpenHHEA.ReasoningXXX: the encapsulated entity alignment reasoning process. Use run function to start entity alignment reasoning.
• OpenHHEA.init_model_pipeline: get the predefined entity alignment pipeline, including the training and reasoning. Use run to start the whole entity alignment pipeline.

Usage

For the detailed usage of OpenHHEA, including the methods for using customized reasoning workflow and using redefined reasoning workflow, please see example_code.py