Sentiment Analysis API

Flask API exposing an endpoint for determining the sentiment of text.

How to run locally. (Terminal)

The following commands require execution in the project root.

1. Create a virtual environment.

python3 -m venv .venv

2. Activate the virtual environment.

. .venv/bin/activate

The following commands assume you have an active virtual environment.

3. Install dependencies. (Using the Makefile)

make install

⚠ The above make target will install the latest, unpinned requirements from the file: requirements.in. If encounter any dependency-related issue during installation or runtime, fall back to use tested, pinned, and stable requirements defined in the file: requirements.txt

• [Optional] Fall back to stable requirements.

pip install -r requirements.txt

4. Run the application.

python app.py

5. Query the api.

Query: I don't like this game

curl \
-X POST \
-H "Content-Type: application/json" \
-d '{"input": "I don't like this game"}' \
"http://127.0.0.1:10010/api/v1/sentiment"

Output:

{
  "prediction": {
    "confidence": 0.9779077172279358,
    "label": "Negative"
  }
}

How to run locally. (Docker)

1. Prerequisite: Make sure you have docker installed. Link

2. Build the docker image. This uses the Docker file in the repository's root.

docker build --build-arg VERSION=SENTIMENT-v1 -t sentiment-api:dev .

3. Run a container using the image as its base.

docker run --rm -p 10010:10010 --name sentiment-api sentiment-api:dev

4. Querying works like the previous one. You can also use Postman.

5. More Examples:

Query: I like this game

curl \
-X POST \
-H "Content-Type: application/json" \
-d '{"input": "I like this game"}' \
"http://127.0.0.1:10010/api/v1/sentiment"

{
  "prediction": {
    "confidence": 0.9681828022003174,
    "label": "Positive"
  }
}

Query: I have no opinion

curl \
-X POST \
-H "Content-Type: application/json" \
-d '{"input": "I have no opinion"}' \
"http://127.0.0.1:10010/api/v1/sentiment"

{
  "prediction": {
    "confidence": 0.49172118306159973,
    "label": "Neutral"
  }
}

How to run on cloud. (AWS)

1. Code available in branches deploy for AWS Lightsail and ebs for AWS Elasticbeanstalk.
2. Documentations are available via this link, Pages 8 & 9.

Built with ❤️ by martinoywa.