Computer Science > Computation and Language
[Submitted on 30 Aug 2021 (this version), latest version 2 Jul 2022 (v3)]
Title:Towards Consistent Document-level Entity Linking: Joint Models for Entity Linking and Coreference Resolution
View PDFAbstract:We consider the task of document-level entity linking (EL), where it is important to make consistent decisions for entity mentions over the full document jointly. We aim to leverage explicit "connections" among mentions within the document itself: we propose to join the EL task with that of coreference resolution (coref). This is complementary to related works that exploit either (i) implicit document information (e.g., latent relations among entity mentions, or general language models) or (ii) connections between the candidate links (e.g, as inferred from the external knowledge base). Specifically, we cluster mentions that are linked via coreference, and enforce a single EL for all of the clustered mentions together. The latter constraint has the added benefit of increased coverage by joining EL candidate lists for the thus clustered mentions. We formulate the coref+EL problem as a structured prediction task over directed trees and use a globally normalized model to solve it. Experimental results on two datasets show a boost of up to +5% F1-score on both coref and EL tasks, compared to their standalone counterparts. For a subset of hard cases, with individual mentions lacking the correct EL in their candidate entity list, we obtain a +50% increase in accuracy.
Submission history
From: Klim Zaporojets [view email][v1] Mon, 30 Aug 2021 21:46:12 UTC (1,155 KB)
[v2] Mon, 27 Jun 2022 14:37:48 UTC (1,325 KB)
[v3] Sat, 2 Jul 2022 00:54:37 UTC (1,325 KB)
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.