1 School of Computer Science, Cardiff University, 5 The Parade, Cardiff CF24 3AA, UK
2 School of Biological Sciences, University of Southampton, Bassett Crescent East, Southampton SO16 7PX, UK
3 School of Plant Sciences, The University of Reading, Whiteknights, PO Box 221, Reading, RG6 6AS, UK
Other talks in this meeting describe the weaving of species and specimen-based information services into work-flows to achieve more complex analyses. Also, the unreliability inherent in the use of species names referring to changing taxonomic concepts of species is addressed by standards and databases for recording these concepts. In this talk, we focus on services that attempt to increase the reliability of assembly and use of such biodiversity information systems, by seeking to identify and help resolve conflicts in taxonomic concepts.
We are developing a software application (“Litchi”) which checks a taxonomic database treatment against a set of rules for internal consistency, thus assisting contributors and editors to maintain data quality. As a set of services, it can be relatively easily integrated into larger systems. It is already flexible enough to allow further uses.
It can check two treatments for “taxonomic conflicts”, where the same species can be detected as being treated differently. Users, who may not be professional taxonomists, need to be alerted to existence of such discrepancies. Litchi provides a means to manage these relationships when users wish to manipulate information associated with the treatments. It summarises the overlaps and conflicts by means of a “cross-map” which, after possible manual improvement, describes how each taxon in one treatment corresponds to the taxa in another. Thus a user can be assisted to navigate from a taxon in one treatment to the corresponding taxon or taxa in another treatment, despite possible differences in their names.
This “taxonomic intelligence” can be used in other ways. Scientists requiring a report or analysis combining data from more than one treatment can have this combined data set assembled with more confidence in its integrity. Non-specialist users can be enlightened as to the reasons why these differences have arisen, hopefully reducing their frustration with apparently arcane taxonomic practices.
We discuss ways in which the intelligence of Litchi can be improved and its areas of applicability broadened, for example by considering taxonomic hierarchies and non-taxonomic ontologies for other data domains.