A Smart Shopping System for Modern E-Commerce Applications

In a modern E-commerce system, exact and efficient classification of products is one of the major challenge. The user experience of online purchasing is heavily relies on product classification accuracy as well as. Given the large number of products and viable categories, have developing a new framework automatically for assigning the products to suitable categories at scale is desirable. However, inadequate item descriptions, poor data quality and adversarial noise in training data, can lead to low prediction accuracy when applying the Machine Learning (ML) algorithms. In addition, the various influx of new products uploaded daily and the dynamic nature of categories highlight the need for novel intelligent classification models that can control the cost and time required by human editors. To overcome these kind of issues, Squirrel search War Strategy Optimization (SSWSO)_LeNet replica is introduced. SSWSO is a revolutionary nature-inspired optimization algorithm developed for unconstrained optimization issues. The foraging behavior of southern flying squirrels is examined and quantitatively simulated, considering every aspect of their food hunt to achieve the desired optimum. Squirrel search Algorithm (SSA) provides global optimum solutions with excellent convergence behavior. Meanwhile, War Strategy Optimization (WSA) algorithm includes an adaptive weight mechanism that varies from one solution (soldier) to another and is updated weights depends on the soldier's rank during the update phase. Thus, the combined optimization strategies ensure an efficient balance among the exploration and exploitation stages. By integrating WSO and SSA provides better product classification accuracy and superior performance. Furthermore, the experimental findings showed that the SSWSO_LeNet performed better in accuracy, sensitivity, and specificity, with values of 0.976, 0.877, and 0.857, which is impressive compared to state-of-the-art results.