Named Entity Recognition and Classification

This project is a web-based application that uses four different algorithms (ACE, Flair, BERT + CRF, and BiLSTM + CRF) for named entity recognition and classification. The application provides users with a simple interface for entering text and obtaining information about the named entities in that text.

Getting Started

To use the application, you will need to install the required dependencies and start the local server. Follow these steps:

BackEnd

The requirements for all the 4 algorithms is mentioned below:

1. ACE - The requirement instructions can be found in the path: named-entity-recognition/code/backend/ACE/requirements.txt
2. Bert + CRF - The instructions can be found in the path: named-entity-recognition/code/backend/bert_crf/README.md
3. BiLSTM + CRF - The version instructions can be found in the path: named-entity-recognition/code/backend/seqtag-keras/packages.txt and the python version should be 3.6
4. Flair - The requirement instructions can be found in the path: named-entity-recognition/code/backend/flair/readme.txt

FrontEnd

The requirement instructions can be found in the path: named-entity-recognition/code/frontend/README.md

Usage

Once you have the application up and running, you can enter text in the input field and click the "Submit" button to obtain information about the named entities in the text. The application will display the recognized named entities along with their types (e.g., person, location, organization) and the algorithm that was used to identify them.

Algorithms

The application uses four different algorithms for named entity recognition and classification:

1. ACE: A rule-based algorithm that uses handcrafted patterns to identify named entities.
2. Flair: A deep learning algorithm that uses contextualized word embeddings to identify named entities.
3. BERT + CRF: A combination of the BERT language model and a conditional random field (CRF) for named entity recognition.
4. BiLSTM + CRF: A combination of a bidirectional LSTM neural network and a CRF for named entity recognition.

FrontEnd

The frontend of the application is built using React.js, a popular JavaScript library for building user interfaces. The interface is designed to be simple and intuitive, with a clean and modern look.

Contributing

We welcome contributions from the community. If you would like to contribute to the project, please follow these steps:

1. Fork the repository and create a new branch for your changes.
2. Make your changes and submit a pull request.
3. We will review your changes and merge them if they are a good fit for the project.

Acknowledgements

We would like to thank the creators of the CoNLL-2003 dataset for providing the data that we used to train and evaluate our algorithms. We would also like to thank the developers of the ACE, Flair, and CRF++ libraries for their contributions to the field of natural language processing.