Linux Certif

Toute la documentation sur la certification Linux LPI

TAG.3pm

NAME

TAG - Perl-style objects to represent Tree Adjoining Grammars [TAG]

SYNOPSIS

  use TAG;
     
  # Basic parsing
  use TAG::XML::Parser;
  $parser = TAG::XML::Parser->new ();
  $document = $parser->parsefile( 'filename' );
 
  # Creating new objects
  $document = TAG::Document->new();
  $tree = TAG::Tree->new( name => 'w0tn1pn2', family => 'tn1pn2' );
  $node = TAG::Node->new( cat => 'NP', id => 'NP_0', type => 'subst' );
 
  # Accessing TAG objects
  $trees = $tree->{Trees};
  $tree = $tree->{Trees}{'tn1pn2'}{'w0tn1pn2'};
  $tree_name = $tree->{'name'};
 
 

DESCRIPTION

TAG uses Grove to implement a tree-based object model for representing Tree Adjoining Grammars [TAG]. Following the XTAG description, such a grammar provides elementary trees grouped into families, lemma entries, and lexicon entries.

How To Create a Grove

There are several ways for TAGs to come into being, they can be read from a file or string using a parser, or they can be created by your Perl code using the `"new()"' methods of TAG.

The most common way to build a TAG::Document object is using a parser, for instance TAG::XML::Parser, which is an instance of XML::Parser.

Using TAGs

The properties provided by parsers are available directly using Perl's normal syntax for accessing hashes and arrays. For example, to get the name of a tree:

   $tree_name = $tree->{'name'};
 
 

The following is the minimal set of objects and their properties that you are likely to get from all parsers:

TAG::Document

families

An hash containing the families TAG::Family indexed by family names.

trees

An hash containing the elementary trees TAG::Tree indexed by tree names.

lemma

An hash of hash containing the lemma TAG::Lemma indexed by name and category.

lexicon

An hash containing the lexicon entries TAG::Lexicon indexed by the lexical form.

desc

An arrayref of descriptions (TAG::Desc).

TAG::Family

The Family objects represent families

name

A string, the name of the family

trees

An array ref of tree names

desc

An arrayref of descriptions (TAG::Desc).

TAG::Tree

The Tree objects represent trees.

name

A string, the name of the tree

family

An array ref of strings, the name of the families the tree belongs to.

type

A string in 'initial' and 'auxiliary'

root

A TAG::Node, root node of the tree

desc

An arrayref of descriptions (TAG::Desc).

TAG::Node

A node in a Tree object.

id

A string, the id of the node

cat

A string, the syntactic category of the node

type

A string in 'std', 'subst', 'foot', 'lex', or 'anchor' denoting the type of the node.

adj

A string in 'yes', 'no', or 'strict' denoting adjoining conditions on the node.

lex

A string, the lexical value of the node of type 'lex'

top

Empty or a TAG::Feature object, representing the 'top' argument of the node

bot

Empty or a TAG::Feature object, representing the 'bot' argument of the node

child

An array of TAG::Node objects, the child of the node

TAG::Lemma

A lemma entry:

name

A string, the name of the lemma

cat

A string, the syntactic category of the lemma

anchors

An array of TAG::Anchor objects that indicate which trees may be anchored by the lemma.

desc

An arrayref of descriptions (TAG::Desc).

TAG::Lexicon

A lexicon entry:

lex

A string, the lexical form of the entry

lemma

An arrayref of TAG::LemmaRef objects that indicate which lemma are possible for this lexical entry.

desc

An arrayref of descriptions (TAG::Desc).

TAG::LemmaRef

A reference to a lemma entry TAG::Lemma in a lexicon entry:

name

A string, the name of the lemma

cat

A string, the syntactic category of the lemma

feature

A TAG::Feature object, denoting additional constraints on the lemma.

TAG::Anchor

A possible anchoring for a lemma entry (TAG::Lemma):

family

An arrayref of strings, denoting the anchoring families

coanchors

An arrayref of TAG::Coanchor constraints

Equations

An arrayref of TAG::Equation constraints

TAG::Coanchor

A constraint of co-anchoring in a TAG::Anchor object, stipulating that some node should yield some lexical value.

node_id

A string, denoting the node upon which the constraint holds

lex

An arrayref of TAG::Lex objects, denoting a disjunction of possible lexical values for the node.

TAG::Lex

A lexical value (in a TAG::Coanchor object):

text

A string, denoting the lexical value

TAG::Equation

An equation in a TAG::Anchor object that should hold on the 'top' or 'bot' argument of a node.

node_id

A string, denoting the node upon which the equation should hold.

type

A string in 'top' or 'bot', denoting the argument of the node upon which the equation should hold.

feature

A TAG::Feature object, representing the constraints

TAG::Feature

A set of (feature,value), used in TAG::Equation, TAG::Node, and TAG::LemmaRef objects:

f

A 'feature' indexed hashref of TAG:Comp objects

TAG::Comp

To hold a value

name

String, the name of the component

id

String (possibly empty), for multiple occurrences (variables).

value

Array ref of values, represented by TAG::Val, TAG::Plus, or TAG::Minus objects

TAG::Val

A standard value in a TAG::Feature object:

Text

The value

TAG::Link

** DEPRECATED ** see attribute id of TAG::Comp

TAG::Plus

A '+' value in a TAG::Feature object. No properties.

TAG::Minus

A '-' value in a TAG::Feature object. No properties.

TAG::Desc

A description option that can be attached to a document, a family, a tree, a lemma, or a lexicon entry.

text

A string, the content of the description

METHODS

TAG by itself only provides one method, new(), for creating new TAG objects. There are Data::Grove extension modules that give additional methods for working with TAG objects and new extensions can be created as needed.

$obj = TAG::OBJECT->new( [PROPERTIES] )

`"new"' creates a new TAG object with the type OBJECT, and with the initial PROPERTIES. PROPERTIES may be given as either a list of key-value pairs, a hash, or a TAG object to copy. OBJECT may be any of the objects listed above.

WRITING EXTENSIONS

The class `"TAG"' is the superclass of all classes in the XML::TAG module. `"TAG"' is a subclass of `"Data::Grove"'.

If you create an extension and you want to add a method to all TAG objects, then create that method in the TAG package.

AUTHOR

Eric de la Clergerie, Eric.De_La_Clergerie@inria.fr

SEE ALSO

Grove (3) TAG::XML::Parser (3) TAG::LP (3) TAG::RCG (3)

Extensible Markup Language (XML) <http://www.w3c.org/XML>