Gareth Hughes

An Open Hypermedia Link Service Architecture Supporting Multiple Context Models

Abstract

Open hypermedia concerns itself with finding ways to provide links between information using systems that allow the direct manipulation of links. These systems are generally built in component fashion, a major component will be the Link Service. This thesis extends previous work on building link services for the World Wide Web in order to improve the quality of linking that can be provided. A link service delivers links into Web pages dynamically, this allows the system the opportunity to adapt the links to suit the user and content. The more general form of this adaption is to deliver links in some context. The core of this work is to examine how to generalise the delivery of links in context so that new context models and techniques can be introduced without writing a new link service each time.
This thesis examines the problem of defining models of context and how different models are often incompatible. It introduces a new architecture for exploring methods of providing links in context. The architecture consists of a core infrastructure along with a method for dynamically changing the components of the system concerned with the link model and the computation process for generating the final link to deliver.
The effectiveness of the architecture is explored through the undertaking of a number of projects using real-world data. The limits of the design are explored and the analysis focuses on how successful it is possible to be given the goals set.

------------------------------------------------------------------

This PhD incorporated two small projects for the Post Office Research Group. Details of which are below.

An open toolkit to facilitate knowledge extraction and document analysis

Post Office Research Group Project 2

Report [3.7mb PDF]
This advances the architecture of my PhD and brings in support for a number of emerging standards. The overall aim is to show that instead of writing an essentially dead document in a word processor a living XML based report can be written that allows a user to interact with the analysis work of the report author. See the report for a better explanation.
To design a link system using the new XLink, XPath, Xpointer standards and create links to paragraphs of documents in XML (700 pages of Word I converted to XML, I should add). Then use it to generate new documents on specific topics. Eg 'All the best bits about buying bicycles'. Read 2 paragraphs below for this to make sense. In other words it's all about XSL transforms and linkbases built in XLink.
[February 2001 - June 2001]

Enhancing AKAP Documents with Prioritised Dynamic Links

Post Office Research Group Project 1

Report [3.7mb PDF]
A project done for the Post Office Research Group in which I built the link service envisaged in my mini-thesis future work section. The system is used to deliver links into documents with a dynamically computed priority (shown by a colour) according to an algorithm. The key point of the architecture is that the method used to Resolve the links is built into a self-contained Class that can be switched at will. The aim of the final part of my PhD is to build a variety of applications and Resolvers using this system and evaluate how well the Resolver idea works. The
[October 2000]

The chart

This charts my word count since 29 January 2002 when I started using cvs. Note that I reached full draft in July 2002 and submitted in August 2003. Really.

thanks dan

My PhD...

...describes the designing, building and using of open hypermedia link service which processes contextual links. The novelty comes in the way it decides what links to place in documents depending on some form of context. As trying to define context is like trying to knit fog the architecture is as open as possible and links are resolved to their final form by a plug-in Link Resolver. A resolver comprises the code to decide what links to place where and all of the linkbase data and design. There is no fixed link format at all. For example the system can place links depending on factors such as who the reader is, what the date of the document is or its relation to other documents in a set (see the 2 Post Office projects below).

Whilst writing the review sections of my thesis I caught up on the slightly crazy row surrounding Microsoft Smart Tags. I thought this was something that shouldn't be ignored so, with Les Carr's help, wrote a short paper for Hypertext 2002. Microsoft Smart Tags: support, ignore or condemn them ?

My mini-thesis was finished June 2000. "Introducing relevancy and context into an open hypermedia link service." I'm interested in dynamic link services and resolving links in context. The work for the Post Office Research Group is a direct implementation of the work described in my mini thesis. My PhD plan is here but the most recent work is in the second Post Office Project report.
11 April 2002