Continual Few-shot Knowledge Graph Completion

This repo shows the source code of paper: Continual Few-shot Knowledge Graph Completion. In this paper, we propose a novel framework designed to equip the few-shot model with the ability to learn sequentially from novel relations. Specifically, we introduce innovative strategies at both the data and model levels: data-level rehearsal and model-level modulation to address catastrophic forgetting, alongside multi-view relation augmentation aimed at resolving the issue of insufficient novel relations.

Running the Experiments

Requirements

Install dependencies by

pip install -r requirements.txt

Dataset

We use NELL-One and Wiki-One to test our model. The orginal datasets and pretrain embeddings can be downloaded from xiong's repo. We select what we need to meet our continual few-shot setting, please check cfkgc-datasets for more details.

Training