Client-side proxies

Master's thesis, May 2000

Tomas Viberg 

  Table of contents

Abstract

With the growth of the Internet, information overload has become a problem. Because of the sheer amount of available data, there is a need for tools that can find and transform data into useful information. Existing tools, such as online search engines, directories and more or less specialised portals are popular, but they do not adjust very well to individual needs. This thesis examines an alternative approach - client-side proxies, running on the end-user's local machine. More versatile than the original proxy servers, they have the ability to intercept communication to support information retrieval and adaptation of content.

To establish the benefits and drawbacks of the approach, a number of existing proxies has been compared with each other and with applications that use different techniques to perform similar tasks, such as integrated clients (browsers, newsreaders, etc), client plug-ins or Web services. The results of this two-phased evaluation show that client-side proxies have merits that distinguish them from other content processing applications. The combination of direct and exhaustive access to the content, client independence, support for aggregation of functionality and complete access to the power of the local computer is a strong argument to use client-side proxies for content processing. However, when usability or performance is crucial, other approaches could be better. Client-side proxies introduce greater overhead than other approaches do and they are generally harder to install and configure. Consequently, even if client-side proxies are better, there is the risk that they will only be embraced by more advanced users.

Provided as part of this thesis, Blueberry is a framework for content processing. Building on the evaluation results, this Muffin extension exemplifies how to integrate a consistent user interface with the client application to increase usability while maintaining the independence of the proxy approach. Through high-level data abstraction, it also shows a way to help developers of third-party extensions increase their productivity.


  Table of contents