“COMPREHENSIVE DOCUMENT-LEVEL SENTIMENT ANALYSIS IN LITERARY WORKS USING ADVANCED DEEP LEARNING MODELS"

ABSTRACT

This project, titled "Comprehensive Document-Level Sentiment Analysis in Literary Works Using Advanced Deep Learning Models", aims to develop an advanced system for analyzing sentiment in long-form textual content at the document level. The system extracts text from various document formats, including PDFs and DOC files, and processes the extracted content through multiple deep learning models to classify the sentiment as positive, negative, or neutral. The project incorporates a combination of cutting-edge algorithms: BERT for contextual understanding of text, GRU + CNN for sequence modeling and feature extraction, CNN for additional feature extraction, LDA for topic modeling, and XGBoost for precise sentiment classification. The system is trained on a dataset of Amazon product reviews to ensure robust sentiment prediction. The output will provide a sentiment score for each document, indicating the proportions of positive, negative, and neutral sentiment. This approach has significant potential in applications such as literary analysis, customer feedback evaluation, and sentiment-driven decision-making across various sectors.

Keywords: Sentiment Analysis, Document-Level, Deep Learning, BERT, GRU, CNN, XGBoost, LDA, Text Classification, Literary Works.