Interactive Visualization of Paragraph-Level Metadata to Support Corporate Document Use

In recent years a huge research effort has been devoted to making information on the Web more accessible. One strand concentrates on semantics (the meaning of data); within this is work on data description (via metadata) and research to develop tools for presenting and accessing information through metadata visualisation. The book where this chapter appears is a collection exploring the emerging field bringing research on information visualisation together with research towards realising the ‘semantic web’ (Berners-Lee, 1998-1999).

This invited chapter concerns how the visualisation of metadata at paragraph level can support goal-directed searches within documents. It describes a novel approach to document visualisation, claims are supported through an account of how a functional taxonomy of paragraph-level tags (semantic metadata) can be generated via systematic empirical methods. An interactive tool for document visualisation (GridVis), developed by Weiss-Lijn, which realises the proposals is described; there is extensive discussion of user testing and formal, quantitative evaluation of system performance and how findings from these sources provided the rationale for the novel features of the visualisation tool.

The chapter sets GridVis in the context of related pioneering attempts to visualise other document properties and to support goal-directed navigation within documents. The research reported contributes to the agenda for realising the semantic web, although it can apply to any electronic documents. The work was initially motivated by the needs of large organisations to better exploit information ‘stored’ in documents. The empirical aspects of the work were made possible through collaboration and sponsorship from JSainsbury plc; the principal funding was via a postgraduate training partnership award from the DTI and EPSRC. Early developments for this work were widely disseminated as was a method for testing based on theoretical ‘best case’ performance devised to support this work but widely applicable to information visualisation tools.

Janet McDonnell's research centres on the study of design and other kinds of professional work as it takes place in natural settings where many competing practical pressures are at play. The focus of research is on different means of making practices 'visible' e.g. the methods in use, the decision-making behaviour, how ideas get introduced and established, the informational bases (sources, uses and quality) that individuals and groups rely on, and the influences of assumptions and norms. The overarching themes uniting different projects are an interest in supporting reflective practice, increasing the potential for individuals and groups to learn from experience, enabling user engagement in design, communication of expertise, and understanding the social interactional aspects of collaborative activity.