Canonical Text Services protocol: plans for version 2.0

Version 2.0 of the Canonical Text Services protocol will generalize the protocol's definition to accommodate any valid XML, using only the defined citation scheme to retrieve documents and document fragments. As far as possible, CTS 2.0 will be designed to be upward compatible with version 1.2.

Changes to TextInventory

The following changes to the TextInventory structure are proposed:

• TextInventory to include identification of how to validate a document using James Clark's Namespace Routing Language; perhaps a default value will be TEI P4 using DTDs, but any schema may be used. Specifically, we should consider whether we want to insist that an entire CTS document conform to a single namespace; in which case, we could adopt Clark's validate element as a model for indicating how to validate a CTS document.
• Documentation of citation scheme in TextInventory must include the mapping of citation elements to XPath expression. We might continue to assume that citation values will appear on attributes (since this follows the basic design of XML), and allow either literal or "wild-card" expressions in the citation scheme expression. For example, if the question mark represents the place holder for a citation value, and we had a citation where one hierarchical element was encoded on the n attribute of div elements of type "book", and a second element was encoded on n attributes of l elements, we might map the citation as:

    <citation label="scroll" xpath="/TEI.2/text/body//div[@type = 'book']/@n="?">
      <citation label="line" xpath="/TEI.2/text/body//div[@type='book']//l/@n="?"/>
   </citation>
  

• TextInventory to express full information about encoding of language and script using proposed language-script encoding standards.

Changes to Requests

The following changes to the requests of CTS 1.2 are proposed:

• GetTEIHeader deprecated as of CTS 2.0 (to be made obsolete in a future version of CTS); it should be replaced with GetDocumentMetadata, a generic request returning metadata in the Dublin Core namespace.
• GetValidReff should be modified to allow an optional parameter limiting the request to a section of the document. This parameter should be an XPath expression defining a section of the document according to the syntax defined in the document's TextInventory for the document's citation scheme. A request to get second-level valid references for book 1 of the Iliad might include parameters
  • level=1
  • xpath=/TEI.2/text/body//div[@type='book' and @n = '1']

Other questions for consideration

• CTS has assumed that a single citation scheme will apply to a whole document. Is it desirable to consider allowing different citation schemes to apply to different sections of a document?

Links

• the Canonical Text Services Protocol