Design Challenges and Misconceptions in Neural Sequence Labeling
arXiv:1806.04470
Abstract
We investigate the design challenges of constructing effective and efficient neural sequence labeling systems, by reproducing twelve neural sequence labeling models, which include most of the state-of-the-art structures, and conduct a systematic model comparison on three benchmarks (i.e. NER, Chunking, and POS tagging). Misconceptions and inconsistent conclusions in existing literature are examined and clarified under statistical experiments. In the comparison and analysis process, we reach several practical conclusions which can be useful to practitioners.
Accepted by COLING 2018 (Best Paper Award)
Cited by in corpus (22)
- A Survey on Deep Learning for Named Entity Recognition
- Named Entity Recognition and Classification on Historical Documents: A Survey
- Hierarchically-Refined Label Attention Network for Sequence Labeling
- CAN-NER: Convolutional Attention Network for Chinese Named Entity Recognition
- Neural Modeling for Named Entities and Morphology (NEMO^2)
- Dependency-Aware Named Entity Recognition with Relative and Global Attentions
- Structural Knowledge Distillation: Tractably Distilling Information for Structured Predictor
- NCRF++: An Open-source Neural Sequence Labeling Toolkit
- NeuronBlocks: Building Your NLP DNN Models Like Playing Lego
- Experimental Standards for Deep Learning in Natural Language Processing Research
- More Embeddings, Better Sequence Labelers?
- Named Entity Recognition Only from Word Embeddings
- Incorporating Uncertain Segmentation Information into Chinese NER for Social Media Text
- Semantic Role Labeling as Syntactic Dependency Parsing
- ADEQA: A Question Answer based approach for joint ADE-Suspect Extraction using Sequence-To-Sequence Transformers
- NorNE: Annotating Named Entities for Norwegian
- FlexNER: A Flexible LSTM-CNN Stack Framework for Named Entity Recognition
- Cross-Lingual Dependency Parsing Using Code-Mixed TreeBank
- Method and Dataset Entity Mining in Scientific Literature: A CNN + Bi-LSTM Model with Self-attention
- iobes: A Library for Span-Level Processing
- AIN: Fast and Accurate Sequence Labeling with Approximate Inference Network
- Does Chinese BERT Encode Word Structure?