| Two important aspects of any e-business are (1) providing personalized interaction with a customer, and (2) improving the quality of service that visitors receive at the e-business site from a site performance perspective. This dissertation will propose approaches to address, in part, the two above-mentioned aspects.; 1. Discovering critical edge sequences in e-commerce catalogs . E-business sites allow collection of vast amounts of navigational data. These massive data stores can be used to discover mostly traversed segments (Critical Edge Sequences, CES) within the catalogs. CESs can be used to increase the effectiveness of delivering targeted personalized content like e-coupons and advertisements for doing cross-sell and up-sale to visitors of the site. It can also be used to improve site performance, site management and gather additional knowledge of customer behavior patterns on the site. Traditional web mining strategies to find CESs is expensive in both space and time. In this work we propose an approximate algorithm to discover CESs. The method is both space efficient and accurate, providing a vast reduction in the storage requirement with a minimum impact on accuracy.; 2. Adding performance and scalability to e-business applications . E-business activities are growing rapidly in terms of both use and infrastructure requirements. Unfortunately, demand is outpacing the capacity of the infrastructure, as evidenced by unacceptably long response times and the ever-increasing need to expand capacity.; Caching is a widespread approach to resolving such scalability and performance problems in e-business infrastructures. Within the object-oriented framework of an e-business application, objects can be cached at various layers of the infrastructure. In such a scenario, the decision of object to cache is typically made in an ad-hoc fashion. We propose a systematic model for caching objects in e-business sites. Specifically, our work consists of three parts: (a) a solution of the model; (b) a set of efficient, practical rules of thumb and heuristics that can be easily applied by a site architect; and (c) an analysis of the performance benefit of applying such a model. |
