Movie Sentiment Analysis and Clip Extraction

This project performs sentiment analysis on movie subtitles and provides a way to extract clips based on the analyzed sentiments. It uses a pre-trained sentiment analysis model to classify emotions in subtitles and visualizes the sentiment trends over time.

Features

• Sentiment Analysis: Analyzes subtitles to classify emotions such as joy, sadness, anger, etc.
• Visualization: Plots sentiment scores over time to show emotional trends in the movie.
• Clip Extraction: Enables identification of specific scenes or clips based on sentiment.

Technologies Used

• Python: Core programming language.
• Transformers: For sentiment analysis using the michellejieli/emotion_text_classifier model.
• Pandas: For data manipulation and analysis.
• Matplotlib & Seaborn: For data visualization.
• MoviePy: (Planned) For extracting video clips based on sentiment.

Installation

1. Clone the repository:

   git clone https://github.com/ZeMendes17/kurz.git

2. Run Docker Compose

   docker compose up --build

3. Ensure you have the subtitle dataset (both .csv files) in the dataset/ folder inside kurz/src/kurz and kurz-recommendation-api/src/. The dataset should include a CSV file with subtitles and their corresponding movie IDs. The dataset is: https://www.kaggle.com/datasets/adiamaan/movie-subtitle-dataset.

Example Output

• A line plot showing sentiment scores across the movie's subtitles.
• Sentiment trends that can help identify emotional highs and lows in the movie.

Acknowledgments

• Hugging Face for the sentiment analysis model.
• Open-source libraries like Pandas, Matplotlib, and Seaborn for data processing and visualization.