- Notifications
You must be signed in to change notification settings - Fork1
Named Entity Recognition with 92.5% of F1-Score, developed in Pytorch using PoS embeddings, Word2Vec
andreabac3/NLP-Named-Entity-Recognition
Folders and files
| Name | Name | Last commit message | Last commit date | |
|---|---|---|---|---|
Repository files navigation
The experiments on NER field let me to build the following model.The model made use of two input, one the PoS embeddings (calculated starting from the SpaCy tool) and the Word embeddings (Word2Vec 300d from Google research)I also tested the model using the Conditional Random Field, as shown in the table result below, I got a more higher score without it when I started using the PoS embeddings.
Check theReportand theSlide Presentation for more information.
| Model | Dev F1 (%) | Test F1 (%) |
|---|---|---|
| Word emb (100d) + BiLSTM | 78.02 | 78.73 |
| Word2Vec + BiLSTM | 80.25 | 81.04 |
| Word2Vec + BiLSTM + CRF | 88.12 | 89.76 |
| Pos Emb + W2v + CRF | 92.11 | 92.37 |
| Pos Emb + W2v | 92.20 | 92.66 |
test.sh is a simple bash script. To run it:
conda create -n nlp2020-hw1 python=3.7conda activate nlp2020-hw1pip install -r requirements.txtbash test.sh data/dev.tsv
Actually, you can replacedata/dev.tsv to point to a different file, as far as the target file has the same format.
About
Named Entity Recognition with 92.5% of F1-Score, developed in Pytorch using PoS embeddings, Word2Vec
Topics
Resources
Uh oh!
There was an error while loading.Please reload this page.