A custom NER model for extracting and labeling chemical compounds from text. This model will extract chemical compounds from a piece of text and label it as one of the following 7 classes:

• ALKANE
• ALKENE
• ALKYNE
• ALCOHOL
• ALDEHYDE
• KETONE
• CAROBOXYLIC ACID (C_ACID)

You can access the ChemNER model in HuggingFace https://huggingface.co/victormurcia/en_chemner

The model has also been deployed in a Streamlit application ChemNER App that also performs Named Entity Linking (NEL) via the PubChem API in order to improve the labeling accuracy by removing entities that are not chemical compounds.

The Streamlit app also provides an annotation visualization for the uploaded text that has color coded labels for all the extracted entities.

Finally, the Streamlit app also has a built-in Wikipedia-API connection to allow for quick querying for Wikipedia articles.

This shows the raw labeling done by the model.

This shows the labeling done in the Streamlit App which has been assisted by the PubChem API.

There are a variety of ways to access the model. The model is hosted in HuggingFace here:

https://huggingface.co/victormurcia/en_chemner

You can also install it in your environment using pip as shown below and then load it via one of the 2 methods below:

!pip install https://huggingface.co/victormurcia/en_chemner/resolve/main/en_chemner-any-py3-none-any.whl

import spacy

nlp = spacy.load("en_chemner")

import en_chemner

nlp = en_chemner.load()

Open an issue ticket or send me an email to [email protected]