Folders and files Name Name Last commit message
Last commit date
parent directory
View all files
Combining Spans into Entities: A Neural Two-Stage Approach for Recognizing Discontiguous Entitie Bailin Wang, Wei Lu .
Improve Traditional Sequence Label Method CRF -> a Neural Method
Two-stage: Segment Extraction + Segment Merging
Neural Architectures for Nested NER through Linearization Jana Strakov´a, Milan Straka, Jan Hajiˇc .
Linear Label -> CONLL-like
Sort By Priority
More Earlier, More Higher Priority
More Longer, More Hi Priority
order search
trivial heuristic merger algorithm
A General Framework for Information Extraction using Dynamic Span Graphs Yi Luan, Dave Wadden, Luheng He, Amy Shah, Mari Ostendorf, Hannaneh Hajishirzi .
IE Framework + Dynamic Span Graph
Multi-granularity Represent
Token-level
Span-level(The Word Neighbor of targe word)
Multi-granularity Propagation
Beam Search
To Solve The Relation info only be used in the first LSTM-layer before.
And Solve Pronouns NER Problem.(6.6% improve)
Entity,Relation,and Event Extraction with Contextualized Span Representations David Wadden, Ulme Wennberg, Yi Luan, Hannaneh Hajishirzi .
Using Bert Replace the original representation modules
Approve The Origin Dynamic Span Graph Module + Bert Bert
Nested Named Entity Recognition via Second-best Sequence Learning and Decoding Takashi Shibuya, Eduard Hovy .
Recursive Separate CRF
Add regular to make the outer entity higher priority than inner entity
optime calculate complexity
Query-Based Named Entity Recognition conference Yuxian Meng, Xiaoya Li, Zijun Sun, Jiwei Li .
build a schema which make ner task to a query answered task.
the transfer of NER task can use the prior info of NER.
Multi-Grained Named Entity Recognition Congying Xia, Chenwei Zhang, Tao Yang, Yaliang Li, Nan Du, Xian Wu, Wei Fan, Fenglong Ma, Philip Yu .
two-stage framework
detection + classification
center search
Merge and Label: A novel neural network architecture for nested NER Joseph Fisher, Andreas Vlachos .
Two-stage
boundaries, 0-1
a complexity neural network
static layer + structure layer + update layer
classification
A Unified MRC Framework for Named Entity Recognition Xiaoya Li, Jingrong Feng, Yuxian Meng, Qinghong Han, Fei Wu, Jiwei Li .
change ner task to a query answered task(same as Query-based NER).
and do nested QA to improve the question answered result.
A Boundary-aware Neural Model for Nested Named Entity Recognition Changmeng Zheng, Yi Cai, Jingyun Xu, Ho-fung Leung, Guandong Xu .
Boundary-aware + Classification two task.
Boundary use BIO -> so assume only have nested NER or flat NER.
Classification use the average hidden of the region representation.
evaluation in GENIA.
Knowledge Guided Named Entity Recognition Pratyay Banerjee, Kuntal Kumar Pal, Murthy Devarakonda, Chitta Baral .
short paper.
view NER task as MAQA.
and add Knowledge.
Specially, context = entity types + question + definition + example.
Evaluation in biomedical dataset.
Linguistically Informed Relation Extraction and Neural Architectures for Nested Named Entity Recognition in BioNLP-OST 2019 Pankaj Gupta, Usama Yaseen, Hinrich Schütze .
Recursive NER to deal Nested NER problem.
Multitask to auxiliary loss.
Rank loss + CRF loss.
using KB and Bagging.
many hand-crafted features.
Learning A Unified Named Entity Tagger From Multiple Partially Annotated Corpora For Efficient Adaptation Xiao Huang, Li Dong, Elizabeth Boschee, Nanyun Peng .
combine multi-corpus entity
the entity tag set of every corpus is different, so the combine is challenge.
O can be every entity
maximum the total likelihood of all possible tag sequences consistent with the gold label.
Semi-Supervised Named Entity Recognition with CRF-VAEs Thomas Effland, Michael Collins .
Semi-supervised
Using VAE in NER as the amortized approximation posterior.
joint tag-encoding Transformer architecture leads to an ≈1% improvement in F1.
resolving unlabeled
test in pre-train model
GRN: Gated Relation Network to Enhance Convolutional Neural Network for Named Entity Recognition Hui Chen, Zijia Lin, Guiguang Ding, Jianguang Lou, Yusen Zhang, Borje Karlsson .
Improving Long-term abilities of CNN.
Four Layer
Embed layer - representation layer
CNN - Contextualized Layer
Gated - Relation Layer
can replace to Direct Fused Layer / ATTention
CRF
TENER: Adapting Transformer Encoder for Name Entity Recognition Hang Yan, Bocao Deng, Xiaonan Li, Xipeng Qiu .
improve Transformer preference in NER Task
change Position Encoder -> relative Position Encoder
remove scaled in MultiHeadAttention
also using character-level information
ner and pos when nothing is capitalized Stephen Mayhew, Tatiana Tsygankova, Dan Roth .
solve capitalize problem in NER and POS.
mix cased and uncased sample in training is better way than Truecase.
Pretrained Encyclopedia: Weakly Supervised Knowledge-Pretrained Language Model Wenhan Xiong, Jingfei Du, William Yang Wang, Veselin Stoyanov .
add KB in pretrained process.
solve the multi-token entity predict problem.
random replace entity to same type entity.
verify in zero-shot entity and commonsense QA.
Zero-Shot Relation Extraction via Reading Comprehension Omer Levy, Minjoon Seo, Eunsol Choi, Luke Zettlemoyer .
View NER as a MRC task which good at zero-shot.
I think it due to the representation of entity label which replace the traditional project representation space to one flat.
The question temperate which is to determine the performance is writer by human.
add negative sample which have no correct answer in sentence from other NER dataset.
You can’t perform that action at this time.