Department of Computer Science Vassar College Poughkeepsie NY  USA

Equipe Langue et Dialogue LORIA/CNRS  Vandúuvre-lès-Nancy FRANCE

XML version of the CES DTDs

BETA RELEASE

This is a Beta release of XCES, which instantiates the EAGLES Corpus Encoding Standard (CES) DTDs for linguistic corpora, developed by the Department of Computer Science, Vassar College, and Equipe Langue et Dialogue, LORIA/CNRS. XCES is under development and subject to change.

We are developing documentation to support XCES. However, the existing CES documentation supporting general encoding practices for linguistic corpora and tag usage is largely relevant to the XCES instantiation, and should be consulted.

XCES is under development. Because the XML framework provides us with means to go well beyond the capabilities of SGML, this development is taking several forms: (1) XML support for additional types of annotation and resources, including discourse/dialogue, lexicons, and speech; (2) creation of additional XSLT scripts to perform common operations and trasduce among formats (including different annotation formats); (3) development of a set of XML schemas instantiating an abstract data model for linguistic annotations, together with a hierarchy of derived types for a broad range of annotation types; and (4) creation of a repository of annotation formats for "off the shelf" use or easy modification via the XCES schemas.

DTDs for XCES

• xcesDoc.dtd: Encoding conventions for level 1
• xcesAna.dtd: Encoding conventions for aligned data
• xcesAlign.dtd: Encoding conventions for annotated data
• xheader.elt: The CES header
• xtoken.elt: NOT USED (optional)
• xword.elt: NOT USED (optional)
• xlink.elt: NOT USED (see below)

Usage

        <?xml version="1.0"?>
        <!DOCTYPE cesDoc PUBLIC "-//CES//DTD XML cesDoc//EN"
                                "dtd/xcesDoc.dtd" [
        ]>...

        <?xml version="1.0"?>
        <!DOCTYPE cesDoc PUBLIC "-//CES//DTD XML cesAna//EN"
                                "dtd/xcesAna.dtd" [
        ]>...

        <?xml version="1.0"?>
        <!DOCTYPE cesDoc PUBLIC "-//CES//DTD XML cesAlign//EN"
                                "dtd/xcesAlign.dtd" [
        ]>...

Notes

Language Identification

XLink support

Support for the XLink specification by including the sub-dtd xlink.ent (for simple, extended, locators and arc elements) is under development.

XPointer/Xpath support

We are currently implementing the use  of XPointers and XPaths for locator element types.

• Element xptr : Attributes from and to
• Element link : Attributes fromLoc, toLoc and xtargets
  Element linkGrp : Attributes fromLoc and toLoc

XSL Stylesheets

Usage:

        java -mx64m \
                -Dcom.jclark.xsl.sax.parser=com.jclark.xml.sax.Driver \
                com.jclark.xsl.sax.Driver your-xces-doc.xml xsl/html/cesDoc.xsl

Samples

• first chapter of Orwell's 1984 (English),  using the XSL driver driver.xsl. (taken from the Multext-East Orwell 1984parallel aligned corpus.)

• fake sample results  using the XSL driver driver.xsl on the XML-encoded text blah.txt,.

We are currently working on a set of stylesheets to support the cesAna and cesAlign DTDs.