The aim of this study was to measure the attitudes of public library staff towards the Internet. Opinions of training received by staff for use of the Internet were also recorded and the relationship between attitudes and training was analysed and considered. This was deemed of value at a time when public library staff were about to embark on the largest public library training initiative ever undertaken for Information and Communications Technology (ICT) and the installation ofPCs with Internet access in every public library as part of the People's Network. A mixture of quantitative and qualitative research methods were utilised including a questionnaire which incorporated an amended version of the Technology Acceptance Model completed by more than 900 public library staff, interviews with managers, focus groups with a cross-section of staff and an online bulletin board. The study found that the attitudes of most public library staffwere positive towards using the Internet at work. Negativity towards the Internet related to discomfort with the cultural changes taking place in public libraries as a result ofiCT. Attitudes were found to have an integral role in relation to public library staff's willingness to use the Internet; in particular, perceptions of usefulness were very influential. Helping the public use the Internet was generally regarded as a positive experience although finding the time to assist library users was difficult. Training, support and assistance for use of the Internet was well rated although a minority of respondents had not received any training. Ratings of the usefulness of Internet training were related to perceptions of the usefulness, ease of use and intention to use the Internet at work. The popularity of self-directed learning denoted the increased potential for online learning in the future. In contrast with findings from the literature review, informal learning methods such as on-the-job and cascade training were well rated by staff for use of the Internet. The findings of this study suggest that seemingly throwaway comments deriding a new innovation or practice in the public library sphere cannot be easily dismissed and may point to deeper concerns about change and lay bare negative attitudes. In addition, staff demonstrating pessimistic and unconstructive remarks appear to be influential. More worryingly, these attitudes may mean that staff will not use a new technology in the way that managers, policy makers and funding bodies envisage.

In recent years, it has become increasingly difficult for users to find relevant information within the accessible glut. Research in Information Filtering (IF) tackles this problem through a tailored representation of the user interests, a user profile. Traditionally, IF inherits techniques from the related and more well established domains of Information Retrieval and Text Categorisation. These include, linear profile representations that exclude term dependencies and may only effectively represent a single topic of interest, and linear learning algorithms that achieve a steady profile adaptation pace. We argue that these practices are not attuned to the dynamic nature of user interests. A user may be interested in more than one topic in parallel, and both frequent variations and occasional radical changes of interests are inevitable over time. With our experimental system "Nootropia", we achieve adaptive document filtering with a single, multi-topic user profile. A hierarchical term network that takes into account topical and lexical correlations between terms and identifies topic-subtopic relations between them, is used to represent a user's multiple topics of interest and distinguish between them. A series of non-linear document evaluation functions is then established on the hierarchical network. Experiments using a variation of TREC's routing subtask to test the ability of a single profile to represent two and three topics of interest, reveal the approach's superiority over a linear profile representation. Adaptation of this single, multi-topic profile to a variety of changes in the user interests, is achieved through a process of self-organisation that constantly readjusts the profile stucturally, in response to user feedback. We used virtual users and another variation of TREC's routing subtask to test the profile on two learning and two forgetting tasks. The results clearly indicate the profile's ability to adapt to both frequent variations and radical changes in user interests.

The fields of the Semantic Web and Ubiquitous Computing are both relatively new fields within the discipline of Computer Science. Yet both are growing and have begun to overlap as people demand ever-smaller computers with persistent access to the internet. The Semantic Web has the potential to become a global knowledge store duplicating the information on the Web, albeit in a machine-readable form. Such a knowledge base combined with truly ubiquitous systems could provide a great benefit for humans. But what of personal knowledge? Information is generally of more use when linked to other information. Sometimes this information must be kept private, so integrating personal knowledge with the Semantic Web is not desirable. Instead, it should be possible for a computer system to collect and store private knowledge while also being able to augment it with public knowledge from the Web, all without the need for user effort. This thesis begins with a review of both fields, indicating the points at which they overlap. It describes the need for semantic annotation and various processes through which it may be achieved. A method for annotating a human's life using a combination of personal data collected using an ubiquitous system and public data freely available on the Semantic Web is suggested and conceptually compared to human memory. Context-aware computing is described along with its potential to annotate the life of a human being and the hypothesis that today's technology is able to carry out this task is presented. The work then introduces a portable system for automatically logging contextual data and describes a study which used this system to gather life annotations on one specific individual over the course of two years. The implementation of the system and its use is documented and the data collected is presented and evaluated. Finally the thesis offers the conclusion that one type of contextual data is not enough to answer most questions and that multiple forms of data need to be merged in order to get a useful picture of a person's life. The thesis concludes with a brief look into the future of the Semantic Web and how it has the potential to assist in achieving better results in this field of study.

The volume and variety of information available on the Internet has experienced exponential growth, presenting a difficulty to users wishing to obtain information that accurately matches their interests. A number of factors affect the accuracy of matching user interests and the retrieved documents. First, is the fact that users often do not present queries to information retrieval systems in the form that optimally represents the information they want. Secondly, the measure of a document's relevance is highly subjective and variable between different users. This thesis addresses this problem with an adaptive approach that relies on evolutionary user-modelling. The proposed information retrieval system learns user needs from user-provided relevance feedback. The method combines a qualitative feedback measure obtained using fuzzy inference, and quantitative feedback based on evolutionary algorithms (Genetic Algorithms) fitness measures. Furthermore, the retrieval system's design approach is based on a multi-agent design approach, in order to handle the complexities of the information retrieval system including: document indexing, relevance feedback, user modelling, filtering and ranking the retrieve documents. The major contribution of this research are the combination of genetic algorithms and fuzzy relevance feedback for modelling adaptive behaviour, which is compared against conventional relevance feedback. Novel Genetic Algorithms operators are proposed within the context of textual; the encoding and vector space model for document representation is generalised within the same context.

Relevance Feedback is a technique that helps an Information Retrieval system modify a query in response to relevance judgements provided by the user about individual results dis played after an initial retrieval. This thesis begins by proposing an evaluation framework for measuring the effectiveness of feedback algorithms. The simulation-based method in volves a brute force exploration of the outcome of every possible user action. Starting from an initial state, each available alternative is represented as a traversal along one branch of a user decision tree. The use of the framework is illustrated in two situations---searching on devices with small displays and for web search. Three well known RF algorithms, Rocchio, Robertson/Sparck-Jones (RSJ) and Bayesian, are compared for these applications. For small display devices, the algorithms are evaluated in conjunction with two strate gies for presenting search results: the top-D ranked documents and a document ranking that attempts to maximise information gain from the user's choices. Experimental results in dicate that for RSJ feedback which involves an explicit feature selection policy, the greedy top-D display is more appropriate. For the other two algorithms, the exploratory display that maximises information gain produces better results. A user study was conducted to evaluate the performance of the relevance feedback methods with real users and compare the results with the findings from the tree analysis. This comparison between the simulations and real user behaviour indicates that the Bayesian algorithm, coupled with the sampled display, is the most effective. For web-search, two possible representations for web-pages are considered---the textual content of the page and the anchor text of hyperlinks into this page. Results indicate that there is a significant variation in the upper-bound performance of the three RF algorithms and that the Bayesian algorithm approaches the best possible. The relative performance of the three algorithms differed in the two sets of experiments. All other factors being constant, this difference in effectiveness was attributed to the fact that the datasets used in the two cases were different. Also, at a more general level, a relationship was observed between the performance of the original query and benefits of subsequent relevance feedback. The remainder of the thesis looks at properties that characterise sets of documents with the particular aim of identifying measures that are predictive of future performance of statis tical algorithms on these document sets. The central hypothesis is that a set of points (corresponding to documents) are difficult if they lack structure. Three properties are identified---the clustering tendency, sensitivity to perturbation and the local intrinsic dimensionality. The clustering tendency reflects the presence or absence of natural groupings within the data. Perturbation analysis looks at the sensitivity of the similarity metric to small changes in the input. The correlation present in sets of points is measured by the local intrinsic dimensionality therefore indicating the randomness present in them. These properties are shown to be useful for two tasks, namely, measuring the complexity of text datasets and for query performance prediction.

A retrieval model describes the transformation of a query into a set of documents. The question is: what drives this transformation? For semantic information retrieval type of models this transformation is driven by the content and structure of the semantic models. In this case, Knowledge Organization Systems (KOSs) are the semantic models that encode the meaning employed for monolingual and cross-language retrieval. The focus of this research is the relationship between these meanings’ representations and their role and potential in augmenting existing retrieval models effectiveness. The proposed approach is unique in explicitly interpreting a semantic reference as a pointer to a concept in the semantic model that activates all its linked neighboring concepts. It is in fact the formalization of the information retrieval model and the integration of knowledge resources from the Linguistic Linked Open Data cloud that is distinctive from other approaches. The preprocessing of the semantic model using Formal Concept Analysis enables the extraction of conceptual spaces (formal contexts)that are based on sub-graphs from the original structure of the semantic model. The types of conceptual spaces built in this case are limited by the KOSs structural relations relevant to retrieval: exact match, broader, narrower, and related. They capture the definitional and relational aspects of the concepts in the semantic model. Also, each formal context is assigned an operational role in the flow of processes of the retrieval system enabling a clear path towards the implementations of monolingual and cross-lingual systems. By following this model’s theoretical description in constructing a retrieval system, evaluation results have shown statistically significant results in both monolingual and bilingual settings when no methods for query expansion were used. The test suite was run on the Cross-Language Evaluation Forum Domain Specific 2004-2006 collection with additional extensions to match the specifics of this model.

A vast amount of information is today stored in the form of textual documents, many of which are available online. These documents come from different sources and are of different types. They include newspaper articles, books, corporate reports, encyclopedia entries and research papers. At a semantic level, these documents contain knowledge, which was created by explicitly connecting information and expressing it in the form of a natural language. However, a significant amount of knowledge is not explicitly stated in a single document, yet can be derived or discovered by researching, i.e. accessing, comparing, contrasting and analysing, information from multiple documents. Carrying out this work using traditional search interfaces is tedious due to information overload and the difficulty of formulating queries that would help us to discover information we are not aware of. In order to support this exploratory process, we need to be able to effectively navigate between related pieces of information across documents. While information can be connected using manually curated cross-document links, this approach not only does not scale, but cannot systematically assist us in the discovery of sometimes non-obvious (hidden) relationships. Consequently, there is a need for automatic approaches to link discovery. This work studies how people link content, investigates the properties of different link types, presents new methods for automatic link discovery and designs a system in which link discovery is applied on a collection of millions of documents to improve access to public knowledge.