I needed a way to match phrases using a fuzzy approach to prepare an annotated dataset to train a model. After searching in forums etc., I realized that I was not the only one looking for this to work, so I share my solution here.
Clone the repo:
git clone https://github.com/jackmen/fuzzy_spacy.gitCreate a virtual environment and activate:
python3 -m venv your_env_name
cd your_env_name
source bin/activateInstall requiremets:
cd /your_path/fuzzy_spacy/
pip3 install -r requirements.txtRun jupyter notebook
jupyter notebookFollow the steps presented in the jupyter notebook. To run the PhuzzyMatcher on a batch of documents, check the spacy documentation (https://spacy.io).
The code execution is not fast, but was sufficient for my use-case to annotate documents to train a NER model.