ReLiK: Retrieve and LinK, Fast and Accurate Entity Linking and Relation Extraction on an Academic Budget

31 Jul 2024  ยท  Riccardo Orlando, Pere-Lluis Huguet Cabot, Edoardo Barba, Roberto Navigli ยท

Entity Linking (EL) and Relation Extraction (RE) are fundamental tasks in Natural Language Processing, serving as critical components in a wide range of applications. In this paper, we propose ReLiK, a Retriever-Reader architecture for both EL and RE, where, given an input text, the Retriever module undertakes the identification of candidate entities or relations that could potentially appear within the text. Subsequently, the Reader module is tasked to discern the pertinent retrieved entities or relations and establish their alignment with the corresponding textual spans. Notably, we put forward an innovative input representation that incorporates the candidate entities or relations alongside the text, making it possible to link entities or extract relations in a single forward pass and to fully leverage pre-trained language models contextualization capabilities, in contrast with previous Retriever-Reader-based methods, which require a forward pass for each candidate. Our formulation of EL and RE achieves state-of-the-art performance in both in-domain and out-of-domain benchmarks while using academic budget training and with up to 40x inference speed compared to competitors. Finally, we show how our architecture can be used seamlessly for Information Extraction (cIE), i.e. EL + RE, and setting a new state of the art by employing a shared Reader that simultaneously extracts entities and relations.

PDF Abstract

Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Entity Linking AIDA-CoNLL ReLiK-Large Micro-F1 strong 86.4 # 4
Entity Linking AIDA-CoNLL ReLiK-Base Micro-F1 strong 85.3 # 7
Relation Extraction CoNLL04 ReLiK-Large RE+ Micro F1 78.1 # 1
Entity Linking Derczynski ReLiK-Large Micro-F1 56.3 # 1
Entity Linking Derczynski ReLiK-Base Micro-F1 55.6 # 2
Entity Linking KORE50 ReLiK-Base Micro-F1 68 # 2
Entity Linking KORE50 ReLiK-Large Micro-F1 72.8 # 1
Entity Linking MSNBC ReLiK-Base Micro-F1 72.3 # 6
Entity Linking MSNBC ReLiK-Large Micro-F1 75 # 2
Entity Linking N3-Reuters-128 ReLiK-Large Micro-F1 51.7 # 3
Entity Linking N3-Reuters-128 ReLiK-Base Micro-F1 48.1 # 4
Entity Linking N3-RSS-500 ReLiK-Base Micro-F1 41.6 # 2
Entity Linking N3-RSS-500 ReLiK-Large Micro-F1 43 # 1
Relation Extraction NYT ReLiK-Large F1 95 # 1
Relation Extraction NYT ReLiK-Small F1 94.4 # 3
Relation Extraction NYT ReLiK-Base F1 94.8 # 2
Entity Linking OKE-2015 ReLiK-Large Micro-F1 65.1 # 2
Entity Linking OKE-2015 ReLiK-Base Micro-F1 62.5 # 4
Entity Linking OKE-2016 ReLiK-Base Micro-F1 52.3 # 4
Entity Linking OKE-2016 ReLiK-Large Micro-F1 57.2 # 3
Relation Extraction REBEL Relik-Large Triplet F1 (strict EL) 75.6 # 1
Entity Linking REBEL ReLiK-Large Micro-F1 85.1 # 1

Methods