The XML format used by DGA

The annotated texts are saved in XML format, as representing the standard in data description adopted by the linguistic community as the standard way of representing corpora. Although a standard set of XML tags for syntactic annotation does not exist yet, as is the case for morpho-syntactic annotation (XCES), DGA uses a minimal set of tags inspired by XCES. Thus, the XML files produced by DGA can be easily transformed, by means of XSLT, into XML files which are based on a different vocabulary (tag set) meeting the requirements of the user or being in conformity with a future standard.

In order to illustrate the used set of tags, we present the following fragment of a xml file, representing the annotation of the sentence "John has apples" (see What is DGA).

<s>
   <tok>
      <orth>John</orth>
      <ordno>1</ordno>
      <ctag>Noun</ctag>
      <syn>
         <head>2</head>
         <reltype>Subject</reltype>
      </syn>
   </tok>
   <tok>
      <orth>has</orth>
      <ordno>2</ordno>
      <ctag>Verb</ctag>
      <syn>
         <head>4</head>
         <reltype>Predicate</reltype>
      </syn>
   </tok>
   <tok>
      <orth>apples</orth>
      <ordno>3</ordno>
      <ctag>Noun</ctag>
      <syn>
         <head>2</head>
         <reltype>Object</reltype>
      </syn>
   </tok>
</s>

Each sentence is marked by tag <s> ... </s>. Each word of the sentence, together with all information concerning its annotation, is marked by tag <tok> ... </tok>. Within this tag, the orthographic form, as it occurs in the annotated text, is marked by tag <orth> ... </orth>. Tag <ordno> ... </ordno> indicates the number of the word within the sentence (counting is performed starting from the beginning of the sentence). By means of tag <ctag> ... </ctag> the part of speech is specified, while tag <syn> ... </syn> marks the syntactic information. Within tag <syn> ... </syn> the head word is specified by means of its number within the sentence, this number being marked by tag <head> ... </head>. The type of the dependency relation existing between the two words (the one to which the annotation belongs and the head word) is specified by means of tag <reltype> ... </reltype>.