TY - CHAP

T1 - Entity Extraction from Portuguese Legal Documents Using Distant Supervision

AU - Navarezi, Lucas M.

AU - Sakiyama, Kenzo

AU - Rodrigues, Lucas S.

AU - Robaldo, Caio M.O.

AU - Lobato, Gustavo R.

AU - Vilela, Paulo A.

AU - Matsubara, Edson T.

AU - Fernandes, Eraldo R.

PY - 2022

Y1 - 2022

N2 - Most approaches to role-filler entity extraction (REE) rely on large labeled training corpora in which entity mentions are directly annotated in the input document. In this work, we leverage an existing knowledge base (KB) of entities to perform document-level REE from drug seizure petitions. We propose a system that learns to extract entities from petitions to fill 29 roles of a drug seizure event. Although we have access to a KB covering more than 170 thousand entities and six thousand petitions, such that each entity in the KB is linked to a specific petition, the mentions to an entity within a petition’s text are not annotated. The lack of these annotations brings challenges related to mismatches between entity values in the KB and entity mentions in the documents. Additionally, there are entities with same type or same value. Thus, we propose a distant annotation method to overcome these challenges and automatically label petition documents using the available KB. This annotation method includes a parameter that controls the balance between precision and recall. We also propose a strategy to effectively tune this parameter in order to optimize a given metric. We then train a BERT-based sequence labeling model that learns to identify entity mentions and label them. Our system achieves an F1 score of 78.59 with precision over 82%. We also report ablation studies regarding the distant annotation method.

AB - Most approaches to role-filler entity extraction (REE) rely on large labeled training corpora in which entity mentions are directly annotated in the input document. In this work, we leverage an existing knowledge base (KB) of entities to perform document-level REE from drug seizure petitions. We propose a system that learns to extract entities from petitions to fill 29 roles of a drug seizure event. Although we have access to a KB covering more than 170 thousand entities and six thousand petitions, such that each entity in the KB is linked to a specific petition, the mentions to an entity within a petition’s text are not annotated. The lack of these annotations brings challenges related to mismatches between entity values in the KB and entity mentions in the documents. Additionally, there are entities with same type or same value. Thus, we propose a distant annotation method to overcome these challenges and automatically label petition documents using the available KB. This annotation method includes a parameter that controls the balance between precision and recall. We also propose a strategy to effectively tune this parameter in order to optimize a given metric. We then train a BERT-based sequence labeling model that learns to identify entity mentions and label them. Our system achieves an F1 score of 78.59 with precision over 82%. We also report ablation studies regarding the distant annotation method.

KW - BERT

KW - Entity extraction

KW - NLP

KW - Business informatics

KW - Informatics

UR - http://www.scopus.com/inward/record.url?scp=85127166992&partnerID=8YFLogxK

UR - https://www.mendeley.com/catalogue/88c72c46-972e-34e1-b04c-2b883dd525c5/

U2 - 10.1007/978-3-030-98305-5_16

DO - 10.1007/978-3-030-98305-5_16

M3 - Article in conference proceedings

AN - SCOPUS:85127166992

SN - 978-3-030-98304-8

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 166

EP - 176

BT - Computational Processing of the Portuguese Language

A2 - Pinheiro, Vládia

A2 - Gamallo, Pablo

A2 - Amaro, Raquel

A2 - Scarton, Carolina

A2 - Batista, Fernando

A2 - Silva, Diego

A2 - Magro, Catarina

A2 - Pinto, Hugo

PB - Springer Nature Switzerland AG

CY - Cham

T2 - 15th International Conference on the Computational Processing of Portuguese - PROPOR 2022

Y2 - 21 March 2022 through 23 March 2022

ER -