Movie/Play Script Analysis

This GitHub repository represents the code for the Klingon group at the International Semantic Web Research Summer School 2024. This project performs various analyses on a movie script, including data cleaning, topic modeling, character interaction graph creation, sentence theme classification, and character type classification.

Setup

Prerequisites

• Python 3.9 or higher
• pip for package management

Install Dependencies

1. Clone this repository:

   git clone https://github.com/your-repo/movie-script-analysis.git
   cd movie-script-analysis

2. Install the required Python packages:

   pip install -r requirements.txt

3. Create a .env file in the root directory and add your OpenAI API key:

   OPENAI_API_KEY=your_openai_api_key_here
   

Run the Analysis

1. Place your movie script text file in a desired location.

2. Run the analysis with the following command:

   python main.py --input_file path/to/your/script.txt --output_folder path/to/output_folder

3. To run specific algorithms, use the --algorithms argument:

   python main.py --input_file path/to/your/script.txt --output_folder path/to/output_folder --algorithms data_cleaner topic_modeler

Algorithms

The --algorithms argument accepts the following options:

• all: Run all algorithms (default).
• data_cleaner: Clean the movie script.
• topic_modeler: Perform topic modeling on the sentences.
• interaction_graph: Create and draw the interaction graph.
• sentence_classifier: Classify sentences based on themes using OpenAI API.
• character_classifier: Classify characters based on their types using OpenAI API.

Example Commands

• Run all algorithms (default):

  python main.py --input_file ./data/Script.txt --output_folder ./output --seed 42

• Run specific algorithms (data cleaner and topic modeler):

  python main.py --input_file ./data/Script.txt --output_folder ./output --seed 42 --algorithms data_cleaner topic_modeler

File Descriptions

• data_cleaner.py: Contains the DataCleaner class for cleaning and processing the script.
• topic_modeler.py: Contains the TopicModeler class for performing topic modeling on the sentences.
• interaction_graph.py: Contains the InteractionGraph class for creating and visualizing a network graph of character interactions.
• sentence_classifier.py: Contains the SentenceClassifier class for classifying sentences based on themes using OpenAI API.
• character_classifier.py: Contains the CharacterClassifier class for classifying characters based on their types using OpenAI API.
• main.py: Orchestrates the entire workflow.

Dependencies

Ensure you have the following packages installed:

• openai
• python-dotenv
• tqdm
• pandas
• networkx
• matplotlib
• seaborn
• sentence-transformers
• bertopic
• scikit-learn
• accelerate

These can be installed via the provided requirements.txt file:

pip install -r requirements.txt