PMLTQ-Commands
view release on metacpan or search on metacpan
xt/author/treebanks/treex_test/resources/treex_subschema_a_layer.xml view on Meta::CPAN
<?xml version="1.0" encoding="utf-8"?>
<pml_schema xmlns="http://ufal.mff.cuni.cz/pdt/pml/schema/" version="1.1">
<revision>1.0.0</revision>
<description>Subschema for a-layer trees</description>
<import type="iset.type" schema="treex_subschema_interset.xml"/>
<!-- a-layer representation (merged with m-layer) -->
<type name="a-root.type">
<structure role="#NODE" name="a-root">
<member name="id" role="#ID" as_attribute="1" required="1"><cdata format="ID"/></member>
<member name="giza_scores">
<structure>
<member name="counterpart.rf">
<cdata format="PMLREF"/>
</member>
<member name="therevalue">
<cdata format="any"/>
</member>
<member name="backvalue">
<cdata format="any"/>
</member>
</structure>
</member>
<member name="ptree.rf"><cdata format="PMLREF"/></member>
<member name="s.rf"><cdata format="PMLREF"/></member>
<member name="afun"><constant>AuxS</constant></member>
<member name="ord" role="#ORDER" required="1"><cdata format="nonNegativeInteger"/></member>
<member name="children" role="#CHILDNODES"><list type="a-node.type" ordered="1"/></member>
<!-- Should the sentence be parsed once again?
This is useful when tags has been fixed (with the knowledge of the parse). Popel 23.5.2009 -->
<member name="reparse" type="bool.type"/>
<member name="wild_dump"><cdata format="any"/></member>
</structure>
</type>
<type name="a-node.type">
<structure role="#NODE" name="a-node">
<member name="children" role="#CHILDNODES"><list type="a-node.type" ordered="1"/></member>
<!--************** COVERED BY TREEX API METHODS ************-->
<member name="id" role="#ID" as_attribute="1" required="1"><cdata format="ID"/></member>
<!-- former m-layer attributes (reduced) -->
<member name="form"><cdata format="any"/></member>
<member name="lemma"><cdata format="any"/></member>
<member name="tag"><cdata format="any"/></member>
<!-- detailed list of morphosyntactic features (may correspond to the tag attribute) -->
<member name="iset" type="iset.type"/>
<member name="no_space_after" type="bool.type"/>
<member name="translit"><cdata format="any"/></member>
<member name="gloss"><cdata format="any"/></member>
<!-- former a-layer attributes -->
<member name="ord" role="#ORDER" required="1"><cdata format="nonNegativeInteger"/></member>
<!-- Since we started working with the Universal Dependencies, deprel became our main attribute for the dependency relation label.
Its usage is not restricted to UD though. It is just a general label. UD restrictions apply only if it is used in the UD context.
In contrast, afun is a special case of deprel, useful only for the Prague-style trees. Its value range is pre-determined and
some values imply specific treatment of the trees even in the Treex API.
Finally, conll/deprel is a predecessor of deprel. Its name refers to one particular file format, CoNLL-X. It is deprecated. -->
<member name="deprel"><cdata format="any"/></member>
<member name="afun" type="afun.type" />
<member name="is_member" type="bool.type"/>
<!-- CoNLL shared task attributes (deprel, pos, cpos, feat) are grouped in one structure -->
<member name="conll" type="a-conll.type"/>
<!-- Shared modifiers of coordinations can be distinguished in PDT style
just based on the fact they are hanged on the conjunction (coord. head).
However, in other styles (e.g. Stanford) this attribute might be useful. -->
<member name="is_shared_modifier" type="bool.type"/>
<!-- attributes for conversion to t-tree -->
<member name="edge_to_collapse" type="bool.type"/>
<member name="is_auxiliary" type="bool.type"/>
<!-- ??? -->
<member name="is_parenthesis_root" type="bool.type" />
<!-- clause segmentation -->
<member name="clause_number"><cdata format="any"/></member>
<member name="is_clause_head" type="bool.type"/> <!-- We mean finite clause head, but that's the same as with is_clause_head on t-layer!!! -->
<!--************** NOT YET COVERED BY TREEX API METHODS ************-->
<member name="morphcat" type="czech_morphcat.type"/> <!-- This should be in Treex API, i.e. not redundant with tag -->
<!-- surface coreference -->
<member name="coref" type="a_coreference.type"/>
<!-- link to p-layer -->
<member name="p_terminal.rf"><cdata format="PMLREF"/></member>
<member name="alignment">
<list ordered="0" type="align-links.type"/>
</member>
<!-- tentative -->
<member name="chunks"><list ordered="0"><cdata format="any"/></list>
</member>
<member name="wild_dump"><cdata format="any"/></member>
<member name="iset_dump"><cdata format="any"/></member>
</structure>
</type>
<type name="a-conll.type">
<structure>
<member name="deprel"><cdata format="any"/></member> <!--dependency relation-->
<member name="cpos"><cdata format="any"/></member> <!--coarse-grained part-of-speech tag-->
<member name="pos"><cdata format="any"/></member> <!--fine-grained part-of-speech tag-->
<member name="feat"><cdata format="any"/></member> <!--various other features separated by |-->
</structure>
</type>
<type name="a_coreference.type"> <!-- analytická koreference pro Jirku MÃrovského -->
<structure>
<member name="markable" type="bool.type"/> <!-- a token to be considered as anaphor/antecedent -->
<member name="introduction" type="bool.type"/> <!-- first introduction of the entity to the discourse -->
<member name="anaphor_type"> <cdata format="any"/> </member> <!-- the type of the anaphor - pronoun, proper_noun, demonstrative_noun_phrase, common_noun -->
<member name="target-node.rf"> <cdata format="PMLREF"/> </member>
<member name="type"> <cdata format="any"/> </member>
<member name="src"> <cdata format="any"/> </member>
</structure>
</type>
<type name="afun.type">
<choice>
<value>Pred</value>
<value>Pnom</value>
<value>AuxV</value>
<value>Sb</value>
<value>Obj</value>
<value>Atr</value>
<value>Adv</value>
<value>AtrAdv</value>
<value>AdvAtr</value>
<value>Coord</value>
<value>AtrObj</value>
<value>ObjAtr</value>
<value>AtrAtr</value>
<value>AuxT</value>
<value>AuxR</value>
<value>AuxP</value>
<value>Apos</value>
<value>ExD</value>
<value>AuxC</value>
<value>Atv</value>
<value>AtvV</value>
<value>AuxO</value>
<value>AuxZ</value>
<value>AuxY</value>
<value>AuxG</value>
<value>AuxK</value>
<value>AuxX</value>
<!-- afun for articles/determiners used in en_generate. ptacek 2008 -->
<value>Det</value>
<!-- afun for English articles "a", "an", "the" (used in SEnglishM_to_SEnglishA).
Other determiners (this,each,any,some,every,no... PennTag=DT) have usually afun Atr. Popel 9.5.2009 -->
<value>AuxA</value>
<!-- afun for english negation "not". On PDT-like t-layer it should be #Neg node, so it's not auxiliary, I think. Popel 9.5.2009 -->
<value>Neg</value>
<!-- values added because of Prague Arabic Dependency Treebank -->
<value>AuxM</value>
<value>AuxE</value>
<value>Ante</value>
<value>PredE</value>
<value>PredC</value>
<value>PredM</value>
<value>PredP</value>
<!-- values added because of Tamil Dependency Treebank (TamilTB.v0.1) by Loganathan Ramasamy -->
<value>AAdjn</value>
<value>AComp</value>
<value>AdjAtr</value>
<value>CC</value>
<value>Comp</value>
<!-- values added because of Ancient Greek Dependency Treebank -->
<value>OComp</value> <!--object complement-->
<!-- values added because of other annotation styles -->
<!-- In PDT, apposition is treated as a paratactic structure (similarly to coordinations) governed by a comma with afun=Apos.
In HamleDT, the first member of apposition governs the second member which has afun=Apposition and governs the (optional) comma.
We cannot reuse afun=Apos for HamleDT because it would break several tests and assumptions. Martin Popel 11.10.2012 -->
<value>Apposition</value>
<!-- The following values are needed for non-Prague coordination styles, non-Prague PPs (no AuxP, afun of the whole PP labeled there) etc. -->
<value>CoordArg</value><!-- conjunct, not the first one -->
<value>PrepArg</value><!-- argument of preposition, if preposition has afun of the whole PP, instead of AuxP -->
<value>SubArg</value><!-- argument of subordinating conjunction, if it has afun of the whole clause, instead of AuxC -->
<value>DetArg</value><!-- argument of determiner, if the style dictates that determiners are heads -->
<value>NumArg</value><!-- argument of number, if the style dictates that numbers are heads -->
<value>PossArg</value><!-- argument of possessive expression (indicating the possessor, i.e. the argument is the possessed thing) -->
<value>AdjArg</value><!-- argument of adjective, i.e. this is the modified noun, if the style dictates that adjectives shall govern -->
<!-- afun for unrecognized value -->
<value>NR</value>
</choice>
</type>
<type name="czech_morphcat.type">
<structure>
<member name="pos"><cdata format="any"/></member>
<member name="subpos"><cdata format="any"/></member>
<member name="synpos"><cdata format="any"/></member>
<member name="gender"><cdata format="any"/></member>
<member name="number"><cdata format="any"/></member>
<member name="case"><cdata format="any"/></member>
<member name="possgender"><cdata format="any"/></member>
<member name="possnumber"><cdata format="any"/></member>
<member name="person"><cdata format="any"/></member>
<member name="tense"><cdata format="any"/></member>
<member name="grade"><cdata format="any"/></member>
<member name="negation"><cdata format="any"/></member>
<member name="voice"><cdata format="any"/></member>
<member name="reserve1"><cdata format="any"/></member>
<member name="reserve2"><cdata format="any"/></member>
<!-- ptacek 5.8.2008 for en/cz_generate -->
<member name="compound_person"><cdata format="any"/></member>
<member name="compound_number"><cdata format="any"/></member>
<member name="compound_gender"><cdata format="any"/></member>
<member name="ordering_type"><cdata format="any"/></member>
<!-- zabokrtsky because of Arabic -->
<member name="definiteness"><cdata format="any"/></member>
<member name="mood"><cdata format="any"/></member>
</structure>
</type>
<type name="align-links.type">
<structure>
<member name="counterpart.rf">
<cdata format="PMLREF"/>
</member>
<member name="type">
<cdata format="any"/>
</member>
</structure>
</type>
<type name="m-form_change.type">
<choice>
<value>ctcd</value>
<value>spell</value>
<value>insert</value>
<value>num_normalization</value>
</choice>
</type>
<!-- radeji do common -->
<type name="bool.type">
<choice>
<value>0</value>
<value>1</value>
</choice>
</type>
</pml_schema>
( run in 0.508 second using v1.01-cache-2.11-cpan-39bf76dae61 )