Path-Based Attention Neural Model for Fine-Grained Entity Typing

TL;DR

This paper introduces PAN, an end-to-end neural model that leverages hierarchical type structures and attention mechanisms to improve fine-grained entity typing robustness against label noise.

Contribution

It proposes a novel path-based attention neural model that effectively utilizes hierarchical type information to enhance noise robustness in entity typing.

Findings

PAN outperforms existing methods in experiments

The model effectively handles label noise

Hierarchical structure improves typing accuracy

Abstract

Fine-grained entity typing aims to assign entity mentions in the free text with types arranged in a hierarchical structure. Traditional distant supervision based methods employ a structured data source as a weak supervision and do not need hand-labeled data, but they neglect the label noise in the automatically labeled training corpus. Although recent studies use many features to prune wrong data ahead of training, they suffer from error propagation and bring much complexity. In this paper, we propose an end-to-end typing model, called the path-based attention neural model (PAN), to learn a noise- robust performance by leveraging the hierarchical structure of types. Experiments demonstrate its effectiveness.