Semi-automatic polyclave key generation
Robert A. Morris, Jacob K. Asiedu, Robert D. Stevenson, and Hua Tang, University of Massachusetts at BostonWe describe a system for producing multiple representations of a taxonomic key from an abstract representation. A key author provides a representation of a polyclave key in a data structure we call a "Property List Decorated Graph"(PLD Graph). Typically a tree, this is a graph which has at each node a list of attribute-value pairs from which applications can produce information in a form suitable to themselves or produce queries to databases, the responses to which provide the input to the application's information production. The key can be provided in XML or as an Excel spreadsheet which our software converts to the XML form. At interior nodes, a typical key might have attribute-value pairs that designate diagnostic characters and references to media files that illustrate the choices. Both attributes and values may be fully structured objects. At leaf nodes, a key might have a property that simply denotes a taxon name denoting the identification produced by the path to node, or (more interestingly) it might support a query into a database known to, or discoverable by, an application processing the key. In our use of such keys, these queries typically produce descriptive taxon pages, however an application processing a key is free to do what it wishes with the information in the PLD Graph. From a single PLD Tree authored in Excel, we have produced keys rendered in HTML for the web, in WML for wireless mobile devices, and renderings which generate DigIR queries to our own or other DigIR providers. We have produced such keys for Ithomid butterflies of Costa Rica, for plant families of Costa Rica, and for aquatic invertebrates of eastern Massachusetts and are presently exploring application to identification applications for other disciplines, beginning with neuroscience. Further details are at http://www.cs.umb.edu/efg/keys