LAUSR

As a website evolves to align with users’ changing information needs and interests, its structure can outgrow the original design, accumulating links and pages in unanticipated places. This increases complexity to both web pages and the navigation structure, which could cause difficulty in locating relevant links and information. Though the increasing complexity of website and its impact on users’ psychological perception have been anecdotally well-recognized, the need to address it with a formal and rigorous method is understudied in the literature. This paper is one of the first studies to examine how to streamline website structures to enhance user navigation. We use a widely used metric in the literature - a page's outdegree (the number of links in a page) - as the measurement for complexity, because it not only serves as a good proximity for page complexity but also has a significant implication on website structure. We propose a method based on mathematical programming (MP) model that can significantly reduce users’ cognitive load by effectively eliminating appropriate links from pages with high complexity. We have performed extensive experiments on both a real dataset and very large synthetic datasets with statistical similarities to the real dataset. The results indicate that our method not only significantly reduces web page and structure complexity with very small impact to user navigation, but also can be effectively solved and scales up remarkably well, suggesting it is useful for website maintenance on a progressive basis. In addition, we conduct a study to evaluate the performance of streamlined website structures using the real dataset and the results confirm the validity of our method.