paper-with-me

Papers

Comparing Span Extraction Methods for Semantic Role Labeling

2021-08-01 · ACL (spnlp) 2021 8 · Zhisong Zhang, Emma Strubell, Eduard Hovy

In this work, we empirically compare span extraction methods for the task of semantic role labeling (SRL). While recent progress incorporating pre-trained contextualized representations into neural encoders has greatly improved SRL F1 performance on popular benchmarks, the potential costs and benefits of structured decoding in these models have become less clear. With extensive experiments on PropBank SRL datasets, we find that more structured decoding methods outperform BIO-tagging when using static (word type) embeddings across all experimental settings. However, when used in conjunction with pre-trained contextualized word representations, the benefits are diminished. We also experiment in cross-genre and cross-lingual settings and find similar trends. We further perform speed comparisons and provide analysis on the accuracy-efficiency trade-offs among different decoding methods.

📄 PDF Abstract BibTeX

Code (1)

zzsfornlp/zmsp 공식 구현

Tasks

Semantic Role Labeling

Similar Papers 제목 키워드 기반

From Semantic Roles to Opinion Roles: SRL Data Extraction for Multi-Task and Transfer Learning in Low-Resource ORL

2025-11-11 · Amirmohammad Omidi Galdiani, Sepehr Rezaei Melal, Mohammad Norasteh, Arash Yousefi Jordehi 외 arxiv

This report presents a detailed methodology for constructing a high-quality Semantic Role Labeling (SRL) dataset from the Wall Street Journal (WSJ) portion of the OntoNotes 5.0 corpus and adapting it for Opinion Role Lab…

Semantic Role LabelingTransfer LearningOpinion Mining

OSLAT: Open Set Label Attention Transformer for Medical Entity Retrieval and Span Extraction

2022-07-12 · Raymond Li, Ilya Valmianski, Li Deng, Xavier Amatriain 외

Medical entity span extraction and linking are critical steps for many healthcare NLP tasks. Most existing entity extraction methods either have a fixed vocabulary of medical entities or require span annotations. In this…

Entity LinkingEntity RetrievalRetrieval

Enhancing Document-level Event Argument Extraction with Contextual Clues and Role Relevance

2023-10-08 · Association for Computational Linguistics 2023 7 · Wanlong Liu, Shaohuan Cheng, Dingyi Zeng, Hong Qu

Document-level event argument extraction poses new challenges of long input and cross-sentence inference compared to its sentence-level counterpart. However, most prior works focus on capturing the relations between cand…

Event Argument ExtractionEvent ExtractionSentence

Span-based Joint Entity and Relation Extraction with Attention-based Span-specific and Contextual Semantic Representations

2020-12-01 · COLING 2020 8 · Bin Ji, Jie Yu, Shasha Li, Jun Ma 외

Span-based joint extraction models have shown their efficiency on entity recognition and relation extraction. These models regard text spans as candidate entities and span tuples as candidate relation tuples. Span semant…

Joint Entity and Relation ExtractionRelationRelation Extraction

Entity, Relation, and Event Extraction with Contextualized Span Representations

2019-09-08 · IJCNLP 2019 11 · David Wadden, Ulme Wennberg, Yi Luan, Hannaneh Hajishirzi

We examine the capabilities of a unified, multi-task framework for three information extraction tasks: named entity recognition, relation extraction, and event extraction. Our framework (called DyGIE++) accomplishes all …

Event ExtractionJoint Entity and Relation Extractionnamed-entity-recognitionNamed Entity Recognition+4