Work Weighting Based on User's Browsing History

This chapter presents discussion of word weighting algorithms in user modelling and adaptive information systems. We specifically address two types of user interest: (1) broad and consistent interest; and (2) narrow, spot interest. A user’s consistent interests can be modelled utilising the user’s information access history; a user’s spot interests can be determined based on that. We developed a word-weighting algorithm to measure the user’s spot interest. The information access history of a user is represented as a set of words. It is considered to be a user model. This method weights words in a document according to their relevancy to the user model. The relevancy is measured by the biases of co-occurrence, called the Interest Relevance Measure, between a word in a document and words in the user model. The future methodology of word weighting is described herein while demonstrating our approach.

A Generic Adaptation Framework for Web-Based Hypermedia Systems

This chapter introduces a framework intended for facilitating the implementation of Web-based adaptive hypermedia systems. The framework is orthogonal to Web “serving” approaches, and poses only minimal requirements in that direction. As such, it can be easily integrated into existing, non-adaptive Web-publishing solutions. This chapter presents in detail several aspects of the framework, and provides an overview of its application in the European Commission-funded IST-1999-20656 PALIO project (“Personalised Access to Local Information and Services for Tourists”). Furthermore, it discusses some of the lessons learned from our work on the framework thus far, as well as what we consider the most likely directions of future work in the area.

Personalisation of Advertisements in the Digital TV Context

In this chapter, we discuss personalisation of advertisements in the digital TV environment and propose an effective personalisation approach, taking into account unique domain requirements. The proposed approach combines the widely used Pearson-based collaborative filtering technique, applied on numerical ratings with the user’s lifestyle, a stable characteristic drawn from consumer behaviour theory. We claim that users with similar lifestyles are reliable neighbours and can be utilised for the recommendation of advertisements for any member of their lifestyle neighbourhood. We focus on an inherent limitation of collaborative filtering methods that occurs when few ratings are available for each user and demonstrate that the proposed approach effectively manages this problem. Indeed, the hybrid approach combines the ability of the Pearson-based approach to accommodate rapid changes in user needs and make predictions upon one-click interactions and the advantage of the lifestyle-based approach to handle sparse data, which significantly affects the performance of collaborative filtering prediction methods.

Adaptation Engineering in Adaptive Concept-Based System

This chapter presents main issues and the state of the art of research on adaptation engineering in adaptive concept-based systems. Adaptive concept-based systems are characterised by the prominent role of concept structures, which makes content classification and conceptualisation play central roles in engineering. On top of these concept structures, adaptation is engineered in order to achieve personalisation of both the content and their presentation. For this presentation many systems use hypermedia structures, as that nicely supports the Web-based application. As a consequence, navigation adaptation is also a central issue in system design. Next to modelling domain and adaptation, it is necessary to model the user and what the system knows or assumes of the user. To discuss different approaches to these issues, we have identified three main classes of adaptive concept-based systems. Adaptive Web information systems build the more general class of data-intensive applications. We use the Hera design methodology to explain the properties of this class. The second class of systems is that of adaptive hypermedia systems. On the basis of the AHAM reference model and the AHA! system, we illustrate this class. The third class consists of adaptive task-based systems, for which we present AIMS as a representative.

Evaluating Usage of Dynamic Web Site Structures Through Visualisation

Web-based content is increasingly delivered via dynamic methods. Visualisation tools are required which reveal how users interact with such data structures in order to improve site design and structure, and to form the basis of adaptation rules. Using our DMASC system we describe a method for logging and visualising individual user paths through a database-driven Web site. We outline the visualisation challenges posed in representing dynamic data structures and representations of user movements within those structures. We introduce two new terms to describe approaches to visualising dynamic structures, template structure and served structure. We present a series of maps generated from real usage data and, through these, identify anticipated and unanticipated surf patterns. Through the presentation of case study material, we argue that visualisations are a useful part of good adaptive multimedia strategies and help form user model attributes.

Problems and Pitfalls in the Evaluation of Adaptive Systems

Empirical studies with adaptive systems offer many advantages and opportunities. Nevertheless, there is still a lack of evaluation studies. This chapter lists several problems and pitfalls that arise when evaluating an adaptive system, and provides guidelines and recommendations for workarounds or even avoidance of these problems. Among other things the following issues are covered: relating evaluation studies to the development cycle; saving resources; specifying control conditions, sample, and criteria; asking users for adaptivity effects; reporting results. An overview of existing evaluation frameworks shows which of these problems have been addressed and in which way.

Site Structure and User Navigation

The analysis of the structure of Web sites and patterns of user navigation through these sites is gaining attention from different disciplines, as it enables unobtrusive discovery of user needs. In this chapter we give an overview of models, measures, and methods that can be used for analysis purposes as well as for user-adaptive navigation support. Specific attention is given to the problem of identifying users getting lost. We conclude with a discussion on various personalised navigation aids that benefit from the techniques presented in this chapter.

Designing to Change Users' Information Seeking Behaviour

Hypermedia systems allow information to be created, stored, accessed, and manipulated in a variety of ways. One example of such a system is a digital library (DL). DLs are typically difficult to learn and to use. One aspect of learnability is that novice users should be able to learn how to search effectively; one approach to this is having the system provide context-relevant help. We report on two studies: the first identifies novices’ difficulties, which informed design changes to integrate adaptive help into a DL system; the second illustrates how interface design can influence users’ information seeking behaviour. It focuses on strategies developed and applied by users in response to two types of ‘tips’. This study provides an indication of how the interface can improve inexperienced users’ interactions with DLs and help them develop more sophisticated information seeking strategies, while also creating more adaptive DLs.

Adaptive Virtual Reality Museums on the Web

This chapter presents an architecture for supporting the creation of adaptive virtual reality museums on the Web. It argues whether the task of developing adaptive virtual reality museums is a complex one, presenting key challenges, and should thus be facilitated by means of a supporting architecture and relevant tools. The proposed architecture is flexible enough to cater for a variety of user needs, and modular promoting extensibility, maintainability, and tailorability. Adoption of this architecture will greatly simplify the development of adaptive virtual reality museums, reducing the needed effort to exhibit digitisation and user profile specification; user profiles are further refined dynamically through the user data recorder and the user modelling engine, which provide input for the virtual environment generator.

Information Imperfection as an Inherent Characteristic of Adaptive Hypermedia

daptive hypermedia applications are aimed at tailoring hypermedia structures according to some form of user model, in an attempt to increase the usability and utility of the application for each individual or group. Existing research in the field has resulted in many systems, techniques, and paradigms, both for modelling user data and for the subsequent exploitation of such model for the sake of personalisation. As a matter of fact, the majority of adaptive hypermedia systems work with user models that are imperfect in some way, and the theories or hypotheses that guide adaptation are also often of a heuristic or approximate nature. Although some existing systems provide explicit means for dealing with imperfection in one or several of its multiple facets, there exists a lack of support for information imperfection in adaptive hypermedia models and architectures. In an attempt to provide such conceptual support, the MAZE model was proposed as a generalisation of an existing abstract hypermedia model, providing built-in support for fuzzy set-theoretic notions. This chapter provides an overall account of the MAZE model, along with its rationale, and an overview of a possible instance of a MAZE-based architecture. In addition, the use of MAZE to model common adaptive hypermedia technologies is illustrated through a concrete case study.