Enhancing E-Commerce Trust: A Hybrid Deep Learning and Aspect-Based Approach for Fake Review Detection

Abstract

The rapid expansion of e-commerce platforms has transformed the way consumers make purchasing decisions, with online customer reviews playing a crucial role in influencing buyer behavior. Customers often rely on product ratings and written feedback to evaluate product quality, seller reliability, and overall customer satisfaction before making purchases. However, the increasing presence of false, misleading, or deceptive reviews has become a serious issue, negatively affecting consumer trust and damaging the credibility of online marketplaces. Fake reviews are often generated by malicious users, competitors, or sellers themselves to manipulate product reputation, promote low-quality products, or harm competitors, creating an unfair and unreliable shopping environment.This project, titled “Advancing E-Commerce Authenticity: A Novel Fusion Approach Based on Deep Learning and Aspect Features for Detecting False Reviews,” aims to develop an intelligent and efficient system for identifying deceptive reviews using a hybrid approach that combines deep learning techniques with aspect-based feature extraction. The proposed system focuses on analyzing multiple dimensions of review data, including textual content, reviewer behavior, sentiment patterns, linguistic characteristics, and product-specific aspects to accurately classify reviews as genuine or fake.Traditional machine learning methods often depend heavily on manually selected features and may fail to capture complex contextual relationships within review text. To overcome these limitations, deep learning models such as Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNN) are employed to automatically learn hidden semantic patterns, contextual dependencies, and sequential relationships in textual reviews.