This repository contains code to perform sentiment analysis on text using roBERTa.

• User will input a text, based on which we will predict the sentiment
• Model only preditcs 3 sentiments, Negative, Neutral, and Positive.

• This version of pretrained roBERTa model preditcs 3 sentiments (Negative, Neutral, and Positive) as we need.

• Server Framework : Flask
• Deep learning Framework : Transformers, Pytorch
• Data Manipulation : Numpy, Scipy

Please install python for this project in your pc.

• Step 1 : Create a veritual Environment

Create using any software of your choice. If you use Anaconda use the following command for creating.

conda create -n env_name python=3.8

python 3.8 is the preferred version for this project.

• Step 2 : Download or clone code repository

Now download the code base in local repository. You directly download download or you can use git to download the codes if you have git cli installed in your computer.

git clone https://github.com/ibrahim-601/sentiment_analysis_api.git

Then take your terminal to required directory.

• Step 3 : Dependency Installation

To install all dependency required for this project use the following command.

pip install -r requirements.txt

• Step 4 : Run the Project

To run the project use the following command

python app.py

This command will launch a development server in your computer at the following ip 127.0.0.1:5000

• path : 127.0.0.1:5000/analyze
• method : POST
• querystring parameter : text
  • example : text="I really like this API."

N.B: Please use this fixed format for calling this API.

• Reponse : {"sentiment": "positive/negative/neutral"} You will receive the sentiment for the text in the response.

• We need to process the data accordingly before feeding into the model.

• analyze_sentiment() function in the sentiment_analyzer.py does the processing required.

• use sentiment_analyzer.py script for single inference provive the appropiate value in the analyze_sentiment() function.