In-a-rush Sentiment Analysis service using Hugging Face 🤗 and FastAPI. The model used is a Bert-based sentiment analysis model. 🧠

Installation

Using Docker

Install Docker 🐳

Build the docker image:

docker build -t sentiment-container .

Run the docker container with the service. The container might take some time to download the models before starting serving.

docker run -dp 8000:8000 sentiment-container

Locally

Install conda and pytorch accordingly to your OS and hardware

Create the environment

conda create -n sentiment_analysis python=3.8 pip

Activate the environment

conda activate sentiment_analysis

Install requirements

pip install -r requirements.txt

Run the server

cd src
uvicorn main:app --port 8000

Usage

To know more about other endpoints, consult the OpenAPI documentation in http://127.0.0.1:8000/redoc or try out the swagger interactive documentation in http://127.0.0.1:8000/docs

• Sentiment analysis endpoint

Request

Method	POST
Path	/sentiment
Body	{"query_string": "String"}

Response

Code	Body	Description
200	{"label": "Sentiment label", "score": "Sentiment score(confidence)"}	Response with the sentiment label POSITIVE, NEGATIVE and confidence level
422	{"detail": {"loc": [], "msg": "Message", "type": "Type"}}	Validation error with calidation error details.

Sending request using curl

Replace the query_string parameter with the sentence you want to

curl -X 'POST' \
  'http://127.0.0.1:8000/sentiment' \
  -H 'accept: application/json' \
  -H 'Content-Type: application/json' \
  -d '{
  "query_string": "You are my best friend"
}'

You will get a response similar to the following

{
  "label": "POSITIVE",
  "score": 0.9998573064804077
}