ShownetAI

Analyze Your Favorite Series with NLP

In this project, we analyze TV series using advanced Natural Language Processing (NLP) and Large Language Models (LLMs). We start by scraping a custom dataset, then move to building several models to classify themes, create a character network, and develop a chatbot that interacts with characters from the series. Finally, all of this is packaged into a user-friendly web interface using Gradio.

Overview

This project contains five main components, each housed in its respective folder:

• crawler: Web scraping using Scrapy to build a dataset.
• character_network: Create a character relationship network using SpaCy’s NER, NetworkX, and PyViz.
• text_classifier: Train a custom text classifier using LLMs.
• theme_classifier: Extract the main themes of the series using Zero-shot classifiers.
• character_chat_bot: Develop a chatbot to chat with characters using LLMs.

Tech Stack

• Scrapy: For web scraping and dataset collection.
• SpaCy: For Named Entity Recognition (NER).
• NetworkX & PyViz: For building and visualizing character networks.
• Hugging Face Transformers: For implementing LLM-based theme and text classifiers.
• Gradio: For creating a web-based GUI for seamless interaction.

Requirements

Before running the project, install the necessary dependencies by running:

pip install -r requirements.txt