https://github.com/agrawal-rohit/tweet-sense

Sentiment analysis dashboard for Twitter hashtags
https://github.com/agrawal-rohit/tweet-sense

cnn lstm sentiment-analysis tweepy twitter-api twitter-sentiment-analysis

Last synced: 21 days ago
JSON representation

Sentiment analysis dashboard for Twitter hashtags

• Host: GitHub
• URL: https://github.com/agrawal-rohit/tweet-sense
• Owner: agrawal-rohit
• Created: 2019-06-10T13:37:59.000Z (about 5 years ago)
• Default Branch: master
• Last Pushed: 2024-03-29T15:11:16.000Z (3 months ago)
• Last Synced: 2024-05-01T15:39:40.489Z (about 2 months ago)
• Topics: cnn, lstm, sentiment-analysis, tweepy, twitter-api, twitter-sentiment-analysis
• Language: Jupyter Notebook
• Homepage:
• Size: 14.5 MB
• Stars: 100
• Watchers: 2
• Forks: 34
• Open Issues: 5
• Metadata Files:
  • Readme: README.md
  • Funding: FUNDING.yml

Lists

• my-awesome-stars - agrawal-rohit/tweet-sense - Sentiment analysis dashboard for Twitter hashtags (Jupyter Notebook)

README

        
![App](imgs/demo.gif)

This web application allows users to analyze sentiments across Twitter hashtags/terms. It's built using React and Django, leveraging an LSTM model trained on the [Kaggle Sentiment140 dataset](https://www.kaggle.com/kazanova/sentiment140). The model is served as a REST API to the ReactJS frontend.

## Features

- **Sentiment Analysis:** Using an LSTM model to analyze sentiments on Twitter.

- **Integration with [Tweepy](https://www.tweepy.org/):** For fetching real-time tweets.

- **Wikipedia API:** To provide context about the hashtags.

- **Visual Examples:** Displaying tweets with their predicted sentiments.

- **Additional Resource:** A kernel for sentiment classification using CNN + 1D pooling is available [here](https://www.kaggle.com/thatawkwardguy/twitter-sentiment-classification-using-cnns).

![Untitled Diagram (6)](https://user-images.githubusercontent.com/29514438/59569258-5f55b700-90a4-11e9-8167-60f53a765c02.jpg)

## Getting Started

### Prerequisites

- Docker installed on your system.

- Twitter API Bearer Token from [Twitter Developer Portal](https://developer.twitter.com/en/portal/projects-and-apps).

### Running the Application

1. **Model Setup:**

    - Download the [trained CNN model](https://drive.google.com/file/d/1ckK5m4JysFKtBuC9yCnEaHe6cxOgXlG8/view?usp=sharing) and place it in the `server/main` folder. *(Note: To use the LSTM model, follow the training steps below and save the model in the `server/main` folder. Modify the loaded model name in `server/main/init.py`.)*

2. **Configuration:**

    - Add your Twitter API Bearer Token to `server/main/config.py`.

3. **Starting the App:**

    - Run `docker-compose up --build` in the terminal from the root directory.

    - Access the app via http://localhost:3000.

### Training the Model

#### CNN Model

Run the [Kaggle Notebook for CNN Sentiment Classification](https://www.kaggle.com/thatawkwardguy/twitter-sentiment-classification-using-cnns).

#### LSTM Model

- Download the [Kaggle Sentiment140 dataset](https://www.kaggle.com/kazanova/sentiment140) and place it as `sentiment140.csv` in the root folder.

- Execute the code in `Twitter Sentiment Analysis.ipynb`.

*(Note: The LSTM model requires more time to train due to its sequential nature. It offers performance similar to the CNN model, but a GPU is recommended for faster processing.)*