Multi-criteria website optimization using multi-objective ACO

The rapid growth of Internet has led to an unprecedented rise to e-commerce. Organizations, while exploiting this opportunity, at the same time are themselves facing stiff competition in order to sustain and rise in this dynamic online market. They are investing a lot for improving their online presence through an effective website design. Such website designs need to address the two-fold challenge of improving the user's navigation experience, even while simultaneously increasing the turnover of the organization. Optimal configuration of web items entails optimization of key criteria like minimization of download time, maximization of visualization and maximizing the potential sale of products or services available through the underlying website configuration, among others. This multi-criteria website optimization (MCWSO) problem has already been formulated as an aggregated weighted sum of the three objectives and solved using the genetic algorithm (GA). It is impractical to have aprior knowledge of the weights for the three objectives, as varied classes of users have different preferences for different criteria. Thus, there is a need to simultaneously optimize the three objectives in order to achieve trade-off solutions, having wider spreads on the Pareto front. Accordingly in this paper, a Pareto based multi-objective ant colony optimization (ACO) based MCWSO algorithm, that achieves such trade-off solutions, has been proposed. Experimental results show that the multi-objective ACO based MCWSO algorithm, in comparison to the GA based MCWSO algorithm, is able to generate Top-K web object sequences that are capable of catering to varied classes of users.