Hugging Face's logo

Spaces:
FusioNER
/
README
Configuration error

App Files Community
README / README.md
etzion's picture
etzion
Update README.md
78a7ae3 verified
preview code
|
raw
history blame contribute delete
14 kB

Fusion NER Models

Here you can find NER models for Fusion project!

Table of content:

  1. NER-Models
  2. Results
  3. Hebrew NLP models
  4. Footnotes

NER Models:

Here you can find a description on each of our models. Each row contains the model nickname, training description, model path (LINK), source dataset (with LINK), base model and entity types.

model name model description model path datasets link to dataset base model entity types trainer
Basic Basic training on IAHALT FusioNER/Basic_IAHALT IAHALT FusioNER/Basic HeRo classic[4] Etzion
Vitaly Vitaly training on IAHALT (with BI-BI problem) FusioNER/Vitaly_NER IAHALT FusioNER/Vitaly HeRo classic[4] Vitaly
Name-Sentences Training on IAHALT + Name-Sentences FusioNER/Name-Sentences IAHALT FusioNER/Name_Sentences HeRo classic[4] Etzion
Entity-Injection Training on IAHALT + Entity-Injection FusioNER/Entity-Injection IAHALT FusioNER/Entity_Injection HeRo classic[4] Etzion
Smart_Injection Training on IAHALT + Name-Sentences + Entity-Injection FusioNER/Smart_Injection IAHALT FusioNER/Smart_Injection HeRo classic[4] Etzion
NEMO Basic training on NEMO dataset FusioNER/Nemo NEMO FusioNER/NEMO HeRo classic[4] Etzion
IAHALT_and_NEMO Basic training on IAHALT + NEMO FusioNER/IAHALT_and_NEMO IAHALT + NEMO FusioNER/IAHALT_and_NEMO HeRo classic[4] Etzion
IAHALT_and_NEMO_PP Training on IAHALT + NEMO + Name-Sentences + Entity-Injection FusioNER/IAHALT_and_NEMO_and_PP IAHALT + NEMO FusioNER/IAHALT_and_NEMO_PP HeRo classic[4] Etzion
Animals Training on IAHALT + Entity-Injection (of animals names as PER entities) FusioNER/Animals IAHALT FusioNER/Animals HeRo classic[4] Etzion
PRS-Injection Training on IAHALT + Entity-Injection (of PRS names as PER entities) FusioNER/PRS-Injection IAHALT FusioNER/PRS_locations HeRo classic[4] Etzion
DICTA_Basic Training the DICTA model on the basic IAHALT dataset FusioNER/Dicta_Small_Basic IAHALT FusioNER/Smart_Injection DICTA classic[4] Etzion
DICTA_Small_Smart Training the DICTA model on IAHALT + Name-Sentences + Entity-Injection] dataset FusioNER/Dicta_Small_Smart IAHALT FusioNER/Smart_Injection DICTA classic[4] Etzion
DICTA_basic_NER Training the DICTA-ner model on the basic IAHALT dataset FusioNER/DICTA_basic IAHALT FusioNER/Basic DICTA-ner classic[4] Etzion
DICTA_smart_NER Training the DICTA-ner model on IAHALT + Name-Sentences + Entity-Injection] dataset FusioNER/DICTA_Smart IAHALT FusioNER/Smart_Injection DICTA-ner classic[4] Etzion
DICTA_Large_Smart Training the DICTA Large model on IAHALT + Name-Sentences + Entity-Injection] dataset FusioNER/Dicta_Large_Smart IAHALT FusioNER/Smart_Injection DICTA Large classic[4] Etzion
TEC_NER Basic technology NER model FusioNER/tec_ner TEC_NER FusioNER/tec_ner base model TEC Yehoshua

Results

We test our models on the IAHALT test set. We also check another models, such as DictaBert and HeBert. This is the performence results:

Model name Precision Recall F1 - Score Time (in seconds)
IAHALT_and_NEMO_PP 0.714 0.353 0.461 83.128
HeBert 0.574 0.474 0.494 86.483
NEMO 0.553 0.51 0.525 81.422
IAHALT_and_NEMO 0.692 0.678 0.684 83.702
Vitaly 0.883 0.794 0.836 83.773
DictaBert 0.916 0.834 0.872 70.465
DICTA_large 0.917 0.845 0.879 206.251
Name-Sentences 0.895 0.865 0.879 82.674
Basic 0.897 0.866 0.881 84.479
Smart_Injection 0.898 0.867 0.881 82.253
DICTA_Basic 0.903 0.875 0.888 69.419
DICTA_Large_Smart 0.904 0.875 0.889 204.324
DICTA_Small_Smart 0.904 0.875 0.889 70.29

According to the results, we recommend to use DICTA_Small_Smart model.

Hebrew NLP models

You can find in the table Hebrew NLP models:

Model name Link Creator
HeNLP/HeRo https://huggingface.co/HeNLP/HeRo Vitaly Shalumov and Harel Haskey
dicta-il/dictabert https://huggingface.co/dicta-il/dictabert Shaltiel Shmidman and Avi Shmidman and Moshe Koppel
dicta-il/dictabert-large https://huggingface.co/dicta-il/dictabert-large Shaltiel Shmidman and Avi Shmidman and Moshe Koppel
avichr/heBERT https://huggingface.co/avichr/heBERT Avihay Chriqui and Inbal Yahav

Footnotes

[1] Name-Sentences:

Adding to the corpus sentences that contain only the entity we want the network to learn.

[2] Entity-Injection:

Replace a tagged entity in the original corpus with a new entity. By using, this method, the model can learn new entities (not labels!) which the model not extracted before.

[3] BI-BI Problem:

Building training corpus when entities from the same type appear in sequence, labeled as continuations of one another. For example, the text "讛讗专讬 驻讜讟专 讜专讜谉 讜讜讬讝诇讬" would tagged as SINGLE entity. That problem prevent the model to extract entities correctly.

[4] Classic:

The classic NER types:

entity type full name examples
PER Person 讗讚讜诇祝 讛讬讟诇专, 专讜讚讜诇祝 讛住, 诪专讚讻讬 讗谞讬诇讘讬抓
GPE Geopolitical Entity 讙专诪谞讬讛, 驻讜诇讬谉, 讘专诇讬谉, 讜讜专砖讛
LOC Location 诪讝专讞 讗讬专讜驻讛, 讗讙谉 讛讬诐 讛转讬讻讜谉, 讛讙诇讬诇
FAC Facility 讗讜讜砖讜讜讬抓, 诪讙讚诇讬 讛转讗讜诪讬诐, 谞转讘"讙 2000, 专讞讜讘 拽驻诇谉
ORG Organization 讛诪驻诇讙讛 讛谞讗爪讬转, 讞讘专转 讙讜讙诇, 诪诪砖诇转 讞讜祝 讛砖谞讛讘
TIMEX Time Expression 1945, 砖谞转 1993, 讬讜诐 讛砖讜讗讛, 砖谞讜转 讛-90
EVE Event 讛砖讜讗讛, 诪诇讞诪转 讛注讜诇诐 讛砖谞讬讬讛, 砖诇讟讜谉 讛讗驻专讟讛讬讬讚
TTL Title 驻讬讛专专, 拽讬住专, 诪谞讻"诇
ANG Language 注讘专讬转, 注专讘讬转, 讙专诪谞讬转
DUC Product 驻讬讬住讘讜拽, F-16, 转谞讜讘讛
WOA Work of Art 讚讜"讞 诪讘拽专 讛诪讚讬谞讛, 注讬转讜谉 讛讗专抓, 讛讗专讬 驻讜讟专, 转讬拽 2000,
MISC Miscellaneous 拽讜专讜谞讛, 讛转讜 讛讬专讜拽, 诪讚诇讬转 讝讛讘, 讘讬讟拽讜讬谉

Datasets for English NER (for cleaning wrong entities for english texts):

MIT License