Seeking Advice on Named Entity Recognition with AI

Beginners
1

Hello everyone,

I hope I’m in the right place for my question, as I am new to the Hugging Face community. I am currently working on “Named Entity Recognition” in connection with AI. My idea was to use a pre-trained model to extract certain information from a text. However, it turns out that this is quite time-consuming for me, given my limited experience with AI, and I would like to shorten my research process. That’s why I thought I’d ask here for advice on the best approach or model recommendations.

Thank you in advance!

1 Like
2

Hello,

Welcome to the Hugging Face community! For Named Entity Recognition (NER), a great way to speed up your work is by using pre-trained models available in the Hugging Face model hub. If you’re just starting out, I recommend trying models like dbmdz/bert-large-cased-finetuned-conll03-english or dbmdz/bert-large-cased-finetuned-ontonotes as they are fine-tuned on NER tasks and can directly extract entities such as names, locations, and organizations from text.

Here’s a simple example of how to use the model with the Hugging Face transformers library:

from transformers import pipeline

# Load pre-trained NER model
ner_pipeline = pipeline("ner", model="dbmdz/bert-large-cased-finetuned-conll03-english")

# Example text
text = "Hugging Face is based in New York City."

# Run NER
entities = ner_pipeline(text)
print(entities)

This should give you a quick start, and from there, you can fine-tune the model or experiment with different ones depending on your needs.

Good luck with your project! Feel free to reach out if you need further help.

2 Likes
3

Hello Alanturner2, thank you very much for this great answer, it really helps me a lot. I will definitely give it a try soon. One quick question: I mainly work with German texts, so there must be pretrained models for the German language as well, right?

1 Like
4

Yeah, you absolutely right. you can use pre trained German models like google-bert/bert-base-german-dbmdz-uncased, distilbert/distilbert-base-german-cased, FacebookAI/xlm-roberta-large-finetuned-conll03-german, dbmdz/bert-base-german-cased and so on.

1 Like
5

Hello
I have same issue with my AI

1 Like
6

Hello , sorry for the late reply! Thanks again for your answers—I’ll try it as quickly as possible! Should I mark this thread as solved?

1 Like
7

If you mark a question as “resolved”, it will be archived after 12 hours (you can extend this by replying to the question), so it’s better to wait until you’re sure the question has been resolved before marking it as “resolved”.