Biblical Sentiment Analysis

This repository performs sentiment analysis on biblical narratives using multiple Large Language Models (LLMs). The analysis focuses on emotions based on the Ekman emotion model and the GoEmotions emotion model, offering insights into biblical texts from various points of view.

Specifically, the experiment explores sentiment and emotional trajectories within the story of the Prodigal Son, analyzed from the perspectives of the Father, Younger Son, and Older Son. The analysis is conducted using four different LLMs and two distinct text chunking techniques: sliding window chunking and sentence-based chunking.

Features

• Multiple LLMs: Uses four different large language models for comparative analysis.
• Emotion Models: Supports both the Ekman and GoEmotions emotion models.
• Text Chunking: Includes both sliding window and sentence-based chunking techniques.
• Perspective-based Analysis: Sentiment analysis is performed from the viewpoints of three key characters in the Prodigal Son narrative: the Father, Younger Son, and Older Son.

Getting Started

How to Duplicate the Experiment

1. Clone this repository:

   git clone https://github.com/darrellcolehill/biblical-sentiment-analysis.git
   cd biblical-sentiment-analysis

2. Open experiment.ipynb in Jupyter Notebook.

3. Run all cells:

   • From the Jupyter Notebook toolbar, select Cell > Run All to execute the experiment.

The notebook contains preconfigured cells that will guide you through the entire process, from loading the text to visualizing the sentiment and emotional outcomes.

Usage

The notebook processes the Prodigal Son text, analyzes the emotions expressed by each character, and tracks the sentiment progression over time.

LLMs Used

1. joeddav/distilbert-base-uncased-go-emotions-student
2. monologg/bert-base-cased-goemotions-original
3. SamLowe/roberta-base-go_emotions
4. j-hartmann/emotion-english-distilroberta-base

Each LLM provides unique insights based on its architecture and training, allowing for a diverse sentiment and emotion comparison.

Emotion Models Supported

• Ekman Model: Includes six basic emotions—anger, disgust, fear, happiness, sadness, and surprise.
• GoEmotions Model: Includes a broader range of 27 emotions, including neutral, caring, and enthusiasm.

Results

The results can be found in the "results" folder.

Experiment Replication Video