BERT REST API

This repository contains code for a BERT REST API that provides various Natural Language Processing (NLP) functionalities using BERT embeddings. The API includes endpoints for obtaining embeddings, performing sentiment analysis, extracting named entities from input texts and summarising text. A lightweight health check is also exposed for monitoring.

File Structure

The project has the following file structure:

bert-rest-api/
  ├── app/
  │   ├── api.py
  │   ├── model.py
  │   └── requirements.txt
  ├── templates/
  │   └── index.html
  ├── tests/
  │   ├── test_embeddings.py
  │   └── test_api.py
  ├── Dockerfile
  ├── .gitignore
  ├── LICENSE
  └── README.md

• app/: Directory containing the main application code.
  • api.py: Flask application exposing the API.
  • model.py: BERT model utilities.
  • requirements.txt: Runtime Python dependencies.
• templates/: HTML templates for the web interface.
• tests/: Unit tests for key functionality.
• Dockerfile: Docker image configuration.
• .gitignore: Patterns for files ignored by Git.
• LICENSE: Project license.
• README.md: Project documentation and instructions.

Usage

1. Clone the repository:

   git clone https://github.com/abalmumcu/bert-rest-api.git

2. Change to the project directory:

   cd bert-rest-api

3. Install the required dependencies:

   pip install -r app/requirements.txt

4. (Optional) choose a different Hugging Face model by setting the MODEL_NAME environment variable before starting the server. The default is bert-base-uncased.

   export MODEL_NAME=distilbert-base-uncased
   python app/api.py

5. Start the API server:

   python app/api.py

6. Access the API endpoints:

   • Open your web browser and go to http://localhost:8888 to view the home page with API information.

   • Use an API client (e.g., cURL or Postman) to send requests to the endpoints below.

Available endpoints

Method	Endpoint	Description
POST	/api/predict	Sentence embeddings for a list of texts.
POST	/api/word_embeddings	Token-level embeddings for a single text.
POST	/api/sentiment	Sentiment analysis for a list of texts.
POST	/api/entities	Named‑entity recognition for a list of texts.
POST	/api/summarize	Summaries for a list of texts.
GET	/health	Service health information.

Docker

To run the API server inside a Docker container, follow these steps:

1. Build the Docker image:

   docker build -t bert-rest-api .

2. Run the Docker container (override the model with -e MODEL_NAME=... if desired):

   docker run -d -p 8888:8888 bert-rest-api

3. The API server is now running inside the Docker container. Access http://localhost:8888 to see the home page and use the API endpoints as mentioned above.

Tests

Run the test suite with:

pytest

License

This project is licensed under the MIT License.