Effect of depth order on iterative nested named entity recognition models
| Autor: | Xavier Tannier, Yoann Taillé, Perceval Wajsbürt |
|---|---|
| Přispěvatelé: | Laboratoire d'Informatique Médicale et Ingénierie des Connaissances en e-Santé (LIMICS), Institut National de la Santé et de la Recherche Médicale (INSERM)-Sorbonne Université (SU)-Université Sorbonne Paris Nord, Tannier, Xavier |
| Rok vydání: | 2021 |
| Předmět: |
FOS: Computer and information sciences
Computer Science - Machine Learning Computer science named entity recognition computer.software_genre biomedical [INFO.INFO-CL]Computer Science [cs]/Computation and Language [cs.CL] Task (project management) Machine Learning (cs.LG) Set (abstract data type) 03 medical and health sciences Named-entity recognition Depth order 030304 developmental biology Transformer (machine learning model) 0303 health sciences Iterative and incremental development nested entities Computer Science - Computation and Language 030302 biochemistry & molecular biology Biomedical information Order (business) [INFO.INFO-CL] Computer Science [cs]/Computation and Language [cs.CL] Data mining computer Computation and Language (cs.CL) |
| Zdroj: | Conference on Artificial Intelligence in Medecine (AIME 2021) Conference on Artificial Intelligence in Medecine (AIME 2021), Jun 2021, Porto, Portugal HAL Artificial Intelligence in Medicine ISBN: 9783030772109 AIME |
| DOI: | 10.48550/arxiv.2104.01037 |
| Popis: | International audience; This paper studies the effect of the order of depth of mention on nested named entity recognition (NER) models. NER is an essential task in the extraction of biomedical information, and nested entities are common since medical concepts can assemble to form larger entities. Conventional NER systems only predict disjointed entities. Thus, iterative models for nested NER use multiple predictions to enumerate all entities, imposing a predefined order from largest to smallest or smallest to largest. We design an order-agnostic iterative model and a procedure to choose a custom order during training and prediction. To accommodate for this task, we propose a modification of the Transformer architecture to take into account the entities predicted in the previous steps. We provide a set of experiments to study the model's capabilities and the effects of the order on performance. Finally, we show that the smallest to largest order gives the best results. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|