Fine-grained Entity Typing via Label Reasoning

TL;DR

This paper introduces Label Reasoning Network (LRN), a novel method that models label dependencies to improve fine-grained entity typing, achieving state-of-the-art results and addressing long-tail label issues.

Contribution

The paper proposes LRN, which combines auto-regressive and bipartite graph reasoning to better capture label dependencies in entity typing.

Findings

Achieves state-of-the-art performance on ultra fine-grained entity typing benchmarks.

Effectively resolves the long tail label problem.

Models complex label dependencies in an end-to-end manner.

Abstract

Conventional entity typing approaches are based on independent classification paradigms, which make them difficult to recognize inter-dependent, long-tailed and fine-grained entity types. In this paper, we argue that the implicitly entailed extrinsic and intrinsic dependencies between labels can provide critical knowledge to tackle the above challenges. To this end, we propose \emph{Label Reasoning Network(LRN)}, which sequentially reasons fine-grained entity labels by discovering and exploiting label dependencies knowledge entailed in the data. Specifically, LRN utilizes an auto-regressive network to conduct deductive reasoning and a bipartite attribute graph to conduct inductive reasoning between labels, which can effectively model, learn and reason complex label dependencies in a sequence-to-set, end-to-end manner. Experiments show that LRN achieves the state-of-the-art performance on standard ultra fine-grained entity typing benchmarks, and can also resolve the long tail label problem effectively.