Leveraging BERT-GRU for Sentiment-Driven Demand Forecasting & Product Insight in E-Commerce

Methanath, Renjitha (2025) Leveraging BERT-GRU for Sentiment-Driven Demand Forecasting & Product Insight in E-Commerce. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (4MB) | Preview
Preview	PDF (Configuration Manual) Download (312kB) | Preview

Abstract

As e-Commerce continues to expand, getting customer sentiment right has become an imperative requirement for both reliable demand forecasting and inventory management. Traditional methods of forecasting mostly use historical sales data or social media hints alone, with no consideration of the authenticity of reviews and with minimal contextual information. This paper develops a sentiment-based demand forecasting framework by combining a BERT-GRU deep learning model with current state-of-the-art time series forecasting techniques like SARIMAX. Based on supervised Amazon review data, the system identifies fine-grained sentiment, filters out low-quality or artificially created text, and links sentiment polarity, buyer ratings, and review trends to demand forecasting with greater accuracy. The fine-tuned BERT-GRU model achieved an impressive 95% accuracy with consistent performance across all classes. Furthermore, incorporating sentiment as an exogenous variable in SARIMAX yielded a significant improvement in model fit, reducing the AIC from 189 to 181. In closing the loop between actionable forecasting and sentiment insights, this approach addresses concerns like spam reviews and context mismatch, offering a scalable, real-world solution to smarter, optimized decision-making in e-commerce.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Agarwal, Bharat UNSPECIFIED
Uncontrolled Keywords:	Sentiment Analysis; BERT-GRU; Transformer; Review Authenticity; Demand Forecasting; SARIMAX
Subjects:	P Language and Literature > P Philology. Linguistics > Computational linguistics. Natural language processing H Social Sciences > HF Commerce > Marketing > Consumer Behaviour H Social Sciences > HF Commerce > Electronic Commerce Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	26 Aug 2026 08:53
Last Modified:	26 Aug 2026 08:53
URI:	https://norma.ncirl.ie/id/eprint/9644

Actions (login required)

View Item