Logic Mining Approach: Shoppers’ Purchasing Data Extraction via Evolutionary Algorithm

Abstract

Online shopping is a multi-billion-dollar industry worldwide. However, several challenges related to purchase intention can impact the sales of e-commerce. For example, e-commerce platforms are unable to identify which factors contribute to the high sales of a product. Besides, online sellers have difficulty finding products that align with customers’ preferences. Therefore, this work will utilize an artificial neural network to provide knowledge extraction for the online shopping industry or e-commerce platforms that might improve their sales and services. There are limited attempts to propose knowledge extraction with neural network models in the online shopping field, especially research revolving around online shoppers’ purchasing intentions. In this study, 2-satisfiability logic was used to represent the shopping attribute and a special recurrent artificial neural network named Hopfield neural network was employed. In reducing the learning complexity, a genetic algorithm was implemented to optimize the logical rule throughout the learning phase in performing a 2-satisfiability-based reverse analysis method, implemented during the learning phase as this method was compared. The performance of the genetic algorithm with 2-satisfiability-based reverse analysis was measured according to the selected performance evaluation metrics. The simulation suggested that the proposed model outperformed the existing model in doing logic mining for the online shoppers dataset.