A Hybrid Machine Learning and Metaheuristic Based Model for E-Business Risk Management

This study introduces a novel approach to enhance e-business firms by identifying potential risks through the analysis of customers’ Twitter posts. Unlike traditional lexicon-based methods, which can be complex, we propose using machine learning models and metaheuristic optimization techniques for more effective risk detection. Additionally, we employ natural language processing techniques to efficiently handle and analyze text data. This approach aims to assist firms in developing competitive e-business strategies by identifying risks based on customer indications on the platform. To achieve this, we construct three datasets: one for risk detection, one for risk assessment, and one for identifying types of risks, which are used in the training phase. We collected tweets from Amazon customers and found that over a quarter of the sample expressed risks in their tweets, with less than a quarter indicating high risks. The most common types of risks identified were operational and financial risks. Furthermore, various performance measures were calculated to evaluate the effectiveness of the proposed approach. The mean accuracy for the processes of risk detection, risk identification, and risk assessment was found to be 81.66%.