Controlled Terminology Extension

From MircWiki
Revision as of 18:24, 26 September 2006 by Johnperry (talk | contribs)
Jump to navigation Jump to search

This article describes a proposed extension to the MIRCdocument schema for encoding controlled terms.

A MIRCdocument is an XML text stream that organizes teaching file text and resources (images, etc.) in sections roughly corresponding to the way a document would be viewed, with other (typically non-displayed) elements providing additional indexing capabilities. All text in a MIRCdocument is indexed.

The proposed extension is to allow any text in any element to be wrapped in an element that allows it to be indexed as a controlled term, for example:

<keywords> 
    <term 
        scheme-name=”RadLex” 
        term-uri=”termURI”
        href=”referencedResourceURL”
        coords=”x:xCoord, y:yCoord, z:zCoord, t:timeCoord”> 
            word1 
    </term>
     word2 word3
</keywords>

In this example, word1 is a controlled term while word2 and word3 are not. All the attributes of the term element are optional.

The href may be used to associate the term with a specific resource (e.g., image). If href is present, the coords attribute may be used to specify a location in the referenced resource with which the term is associated.

Notes:

To make it easy to insert references to controlled terms in a document, the authoring mechanism must provide access to the terminology. The RSNA MIRC implementation would only provide a RadLex browser, but other authoring applications might provide access to other terminologies.

To support searches on controlled terms, the MIRCquery schema must also be extended. The use cases need to be defined. Three come immediately to mind:

  1. A user might simply indicate that the entire query is to search only controlled terms.
  2. A user might indicate that the entire query is to search only controlled terms from a specific scheme-name.
  3. A user might indicate that a specific term in a query must match only a specific controlled term.

The first two cases can be supported by adding a attributes to the MIRCquery root element. The third would require extending the MIRCquery schema to allow a <term> element within each query field, identifying its contents as being a controlled term. To be a flexible solution, this element would support a term-uri attribute, and to make it possible for the user to supply it, the query service would also have to provide access to the terminology.