E-commerce web application showcase and toolkit with automatic profile moderation system

Abstract

Over the last couple of years a general tendency of a shift towards thinner and more internet oriented applications can be traced. With internet speeds increasing at the current rate, web applications get richer and offer a more responsive user interface. With such prospects, the market nowadays is probably in a much better position to host dot-corn start-ups than during the boom at the end of the previous century. Large web application systems need to be robust, scalable and allow for expansion. The days when dynamic web content was generated by a single CGI script have passed and web sites today are complex systems needing more efforts and experience to develop and maintain than traditional desktop applications. In this dissertation, an attempt to build a similar system, which would later be the foundation for a large scale e-commerce platform, is the first task. Investigation of a robust web programming paradigm and technologies was required with the need to produce a scalable site framework allowing for multi-country and multi-locale content separation. This was achieved by choosing ASP.NET 2.0 development platform on top of which the components such as a URL rewriter module, a dynamic link generator and a hierarchy of custom web section and page classes were built. Additionally, a preview technology codenamed "Atlas" for enriching web user interface was experimentally adopted. Various extensions to ASP .NET framework were built to address the needs of the site. A system attempting to solve the problem of needing human moderation for new customer profiles is yet another part of this dissertation, which utilises inexact string matching techniques packaged in a separate string distance toolkit. As a value added tool and to help the automatic moderation system, a language identification tool based on N-gram frequency profile ranking, was also developed and packaged in a separate library.