PNEL: Pointer Network based End-To-End Entity Linking over Knowledge Graphs

TL;DR

This paper introduces PNEL, an end-to-end entity linking model using Pointer Networks that improves accuracy by eliminating the cascade of errors from span detection, evaluated on Wikidata datasets.

Contribution

The paper presents a novel end-to-end entity linking approach with Pointer Networks, addressing span detection limitations and demonstrating competitive results.

Findings

Achieves competitive performance on Wikidata datasets.

Eliminates cascade errors from span detection in entity linking.

Shows effectiveness of Pointer Networks in end-to-end EL.

Abstract

Question Answering systems are generally modelled as a pipeline consisting of a sequence of steps. In such a pipeline, Entity Linking (EL) is often the first step. Several EL models first perform span detection and then entity disambiguation. In such models errors from the span detection phase cascade to later steps and result in a drop of overall accuracy. Moreover, lack of gold entity spans in training data is a limiting factor for span detector training. Hence the movement towards end-to-end EL models began where no separate span detection step is involved. In this work we present a novel approach to end-to-end EL by applying the popular Pointer Network model, which achieves competitive performance. We demonstrate this in our evaluation over three datasets on the Wikidata Knowledge Graph.