Lexis - emotion detection in text

Lexis is web based application that analyzes given user text input using SVM and shows the emotion(s) in the text. The model is trained on more than 75k tweets, seperated in six categories, based on their hashtag: anger, disgust, fear, happiness, irony, neutral, sadness, sarcasm, surprise.

Hackathon winning project

Lexis was created at a hackathon (HackFMI 8 - hack for data) by me and 5 other university coleagues. The code in this repository is "cleaner" and refactored version of what we presented (the original repository can be seen here). Also here I am using React for UI.

Getting Started

Make sure you have Python3, virtualenv and NodeJS installed on your system.

Clone the repository

git clone https://github.com/frisibeli/lexis-text-analysis.git

Initialize and activate virtualenv

cd lexis-text-analysis/backend
virtualenv -p python3 .
source bin/activate

Installing

Assuming that you have already initialized virtualenv and you are in lexis-text-analysis/backend:

Install python requirements using pip

pip install -r requirements.txt

To run the app execute

python app.py

The app will run on port 8080 by default, but you can change that in app.py Now go to lexis-text-analysis/frontend and execute npm install

To run the webapp execute npm start in the frontend directory.

Demo

Built With

• React - JavaScript library for building user interfaces
• Flask - Flask is a micro web framework written in Python.
• Scikit learn - free software machine learning library for Python