Marking and Exploring Taxonomic Concept Data
Paul Craig, Martin Graham, Jessie Kennedy
Abstract
We will present a summary of our work in developing graphical tools for marking up and exploring relationships between overlapping taxonomies. Two visual tools are presented: the first demonstrates how concept relationship sets between classifications can be constructed using a drag and drop tool, and the latter allows exploration and comparison of multiple, inter-related classifications through these concept relationships.
The first application, Concept Relationship Editor, allows taxonomists to create, edit and delete relationships between pairs of classifications. Users can select a pair of concepts, one per classification, and choose a relationship type to construct between them through a drag and drop metaphor in the user interface.
The second tool, TaxVis, allows the exploration and comparison of multiple such classifications. Comparison of sub-groups or an entire classification can be made against the rest of the classification set either through name matching or through the defined concept relationships, allowing the degree of overlap and the difference between naïve and explicit matching to be observed. Overlapping groups of concepts are indicated using colouring. Explicit relationships for specifically interrogated concepts are drawn as links between the corresponding concept representations in the display. A linked panel showing details of specific concepts and their relationships in text form can also be viewed.
The data sets we use in these prototypes are defined under a subset of the Taxonomic Concept Schema (TCS) TDWG standard, with particular focus on its concept relationship mechanism. The tools are designed to improve accuracy beyond naïve name matching when mapping between related taxonomic classifications. The advantage of using mainly graphic representations to convey such classifications and their inter-relationships is that it allows creating, querying and interpreting results to be performed through point and click operations rather than requiring detailed knowledge of the TCS schema and associated XML mechanisms. The dataset also appears as a cohesive whole rather than a succession of atomic information items as would be returned by a traditional text-based system.
We performed initial usability tests on the prototype applications by asking taxonomists to try simple tasks with the tools, and we are interested in accommodating further user-centred development through empirical testing of the prototypes and associated qualitative feedback.
The first application, Concept Relationship Editor, allows taxonomists to create, edit and delete relationships between pairs of classifications. Users can select a pair of concepts, one per classification, and choose a relationship type to construct between them through a drag and drop metaphor in the user interface.
The second tool, TaxVis, allows the exploration and comparison of multiple such classifications. Comparison of sub-groups or an entire classification can be made against the rest of the classification set either through name matching or through the defined concept relationships, allowing the degree of overlap and the difference between naïve and explicit matching to be observed. Overlapping groups of concepts are indicated using colouring. Explicit relationships for specifically interrogated concepts are drawn as links between the corresponding concept representations in the display. A linked panel showing details of specific concepts and their relationships in text form can also be viewed.
The data sets we use in these prototypes are defined under a subset of the Taxonomic Concept Schema (TCS) TDWG standard, with particular focus on its concept relationship mechanism. The tools are designed to improve accuracy beyond naïve name matching when mapping between related taxonomic classifications. The advantage of using mainly graphic representations to convey such classifications and their inter-relationships is that it allows creating, querying and interpreting results to be performed through point and click operations rather than requiring detailed knowledge of the TCS schema and associated XML mechanisms. The dataset also appears as a cohesive whole rather than a succession of atomic information items as would be returned by a traditional text-based system.
We performed initial usability tests on the prototype applications by asking taxonomists to try simple tasks with the tools, and we are interested in accommodating further user-centred development through empirical testing of the prototypes and associated qualitative feedback.