Exploring the Limits of Few-Shot Link Prediction in Knowledge Graphs

TL;DR

This paper systematically investigates the capabilities and limitations of few-shot link prediction in knowledge graphs, revealing that simple baselines perform surprisingly well and that structural information is limited with few examples.

Contribution

It provides a comprehensive analysis of models in few-shot link prediction, challenging assumptions and highlighting the importance of coarse-grained information over fine-grained structural cues.

Findings

Zero-shot baseline performs strongly despite ignoring relation-specific info.

Few examples limit models to coarse-grained positional information.

Structural information is less useful with very few relation examples.

Abstract

Real-world knowledge graphs are often characterized by low-frequency relations - a challenge that has prompted an increasing interest in few-shot link prediction methods. These methods perform link prediction for a set of new relations, unseen during training, given only a few example facts of each relation at test time. In this work, we perform a systematic study on a spectrum of models derived by generalizing the current state of the art for few-shot link prediction, with the goal of probing the limits of learning in this few-shot setting. We find that a simple zero-shot baseline - which ignores any relation-specific information - achieves surprisingly strong performance. Moreover, experiments on carefully crafted synthetic datasets show that having only a few examples of a relation fundamentally limits models from using fine-grained structural information and only allows for exploiting the coarse-grained positional information of entities. Together, our findings challenge the implicit assumptions and inductive biases of prior work and highlight new directions for research in this area.