GusContini / AWS

AWS Projects

Message Classification with AWS Bedrock and AWS Lambda

This serverless application classifies text messages into predefined categories using the power of Amazon Bedrock's foundation models and AWS Lambda.

Technologies Used

• Amazon Bedrock: Provides access to powerful foundation models (e.g., Llama 2) for natural language processing tasks.

• AWS Lambda: Serverless compute service for running the classification logic.

• AWS API Gateway: Creates a RESTful API endpoint to interact with the Lambda function.

• Boto3: AWS SDK for Python, used to interact with Bedrock and S3.

• Python: Programming language used to implement the Lambda function.

• Amazon S3: Stores the classified messages along with their original content.

Functionality

1. Receive Message: The application receives a POST request containing a message and a list of possible categories.

2. Invoke Bedrock Model: The Lambda function uses the Bedrock API to invoke a foundation model (e.g., Llama 2) with a prompt designed for text classification.

3. Classify Message: The model analyzes the message and returns the most suitable category from the provided list.

4. Save to S3: The original message and the assigned category are saved to an S3 bucket in a text file.

5. Return Response: The API Gateway returns a JSON response with the classified category.

API Endpoint

URL: https://iqzolzrrni.execute-api.us-east-1.amazonaws.com/dev/message-classification

Method: POST

Body (JSON): JSON { "message": "Your message to classify", "categories": ["Category1", "Category2", ...] }

Usage

1. Prepare Your Request:

• Replace "Your message to classify" with the text you want to classify.

• Replace "Category1", "Category2", etc. with your desired categories.

2. Send POST Request: Use a tool like curl, Postman, or your preferred HTTP client to send a POST request to the API endpoint with the JSON body.

3. Receive Response: The API will respond with a JSON object containing the classified_message key and the assigned category as its value.

Example (using curl):

Bash

curl -X POST -H "Content-Type: application/json" -d '{"message":"This is a test message.", "categories":["Sports", "Technology", "Politics"]}' https://iqzolzrrni.execute-api.us-east-1.amazonaws.com/dev/message-classification

Response Example JSON { "classified_message": "Technology" }