LsidLookUp

Introduction

The problem goes like this:

People start labeling the data objects that they own with LSIDs i.e. a herbarium generates LSIDs for its specimens. The herbarium database also has data about TaxonNames? but it isn't authoritative for it and so wants to tag these taxon names with LSIDs from another authority who manages the nomenclature - perhaps IPNI or Index Fungorum.

How does the herbarium tag their data with LSID from some one else?

Jessie Kennedy's team at Napier in Edinburgh looked at this and presented a semantic web based paper at last years TDWG. ZooBank? are talking about implementing such an interfaces as well. We have talked about developing a generic interface definition for looking up LSIDs but have not had resources to pursue it yet.

Basically the interface would involve a client passing a server a partially populated object and the server would return the LSIDs of the objects that best matched that object. The client could then tag the rows in its database either automatically or with some human interaction - possible a mixture of both.

This subject will definitely become and issue as more providers go live.

Thoughts and contributions would be welcome.

Straw Man Interface Definition

• Server advertises endpoint as URL for particular class from TDWG ontology that it will service.
• Client makes HTTP POST request to server that contains a XML RDF instance of the class with some of the fields filled in with its known data.
• Server returns either:
  • an empty RDF graph - there were no matches
  • a graph containing a single object of the required with the correct LSID in it.
  • a graph containing multiple objects of the class each with and LSID.

A property would need to be added to the classes to show the degree of confidence the server had in the match.

• WebHome