Skip to content

Latest commit

 

History

History
190 lines (145 loc) · 4.36 KB

README.md

File metadata and controls

190 lines (145 loc) · 4.36 KB
Raw

Mood Marker API 🃏

The Mood Marker API is a powerful, lightweight Natural Language Processing (NLP) API tailored for seamless integration and deployment on AWS.

It offers a suite of endpoints for conducting essential NLP tasks, including sentiment analysis and Named Entity Recognition (NER), designed to enhance your applications with the capability to understand and interpret the emotional and factual content of text data.

Try It Out!

MoodMarkerAPI is deployed at https://langtool.net

curl --request POST \
  --url https://langtool.net/sentiment \
  --data 'text=Were no strangers to love. You know the rules and so do I (Do I) · Never gonna give you up. Never gonna let you down'

🚀 Analyze Sentiment

To perform sentiment analysis, use the curl command as follows:

curl -X POST http://localhost/sentiment -d "text=this is text to be analyzed"

This endpoint expects a form POST with the text field containing the text you wish to analyze. It returns a detailed JSON object with the original text, a comprehensive VADER sentiment score, and emotion scores to quantify the sentiment of the input text accurately.

Example of a successful response:

{
  "emotion_scores": {
    "anticipation": 1,
    "joy": 3,
    "positive": 5,
    "surprise": 1
  },
  "text": "Despite the grey skies, John and Maria's wedding in Seattle was filled with joy, laughter, and an overwhelming sense of love, truly a heartwarming event.",
  "vader_emotion_scores": {
    "compound": 0.952,
    "neg": 0.031,
    "neu": 0.509,
    "pos": 0.461
  }
}

Errors will be returned in the following format:

{
  "error": "No body found in the request."
}

🚀 Extract Named Entities

To extract named entities from text, use the following curl command:

curl -X POST http://localhost/ner -d "text=this is text to be analyzed"

This endpoint accepts a URL Encoded Form post with the text property set to the text you wish to analyze for entities. It returns a JSON object containing the identified entities, enhancing your text's understanding.

Example response:

{
  "named_entities": [
    {
      "label": "PERSON",
      "text": "John"
    },
    {
      "label": "PERSON",
      "text": "Maria"
    },
    {
      "label": "GPE",
      "text": "Seattle"
    }
  ],
  "text": "Despite the grey skies, John and Maria's wedding in Seattle was filled with joy, laughter, and an overwhelming sense of love, truly a heartwarming event."
}

Errors will follow this format:

{
  "error": "No body found in the request."
}

Noun Phrases

To extract the nouns and their roots:

curl -X POST http://localhost/nouns -d "text=The quick brown fox..."
{
  "noun_phrases": [
    {
      "root_dep": "nsubj",
      "root_text": "fox",
      "text": "The quick brown fox"
    },
    {
      "root_dep": "pobj",
      "root_text": "dog",
      "text": "the lazy dog"
    }
  ],
  "text": "The quick brown fox jumped over the lazy dog"
}

Getting Started 🏠

Clone the repository, then navigate to the root directory.

git clone [email protected]:kmesiab/mood-marker-api.git && \
cd mood-marker-api

To initiate the Mood Marker API and its dependencies on your local machine, execute the following command:

make docker-up

This command spins up the application, making it ready for local development and testing. You can now access the API at http://localhost.

Deploying Docker to AWS ECR 📦

Authenticate with AWS ECR 🔑:

make ecr-auth

Build and deploy the Docker image to AWS 🚀:

make ecr-deploy

Deploying Task to AWS ECS 🌐

cd terraform && tf init && tf apply

See the ./terraform/README.md for more information on the AWS infrastructure deployment.