com.worksap.nlp.sudachi

Class PathRewritePlugin

java.lang.Object

com.worksap.nlp.sudachi.PathRewritePlugin

public abstract class PathRewritePlugin
extends Object

A plugin that rewrite the best path of the lattice.

Dictionary initialize this plugin with Settings. It can be referred as Plugin.settings.

The following is an example of settings.

 
   {
     "class" : "com.worksap.nlp.sudachi.PathRewritePlugin",
     "example" : "example setting"
   }
 
 

Field Summary

Fields

Modifier and Type	Field and Description
protected Settings	settings

Constructor Summary

Constructors

Constructor and Description
PathRewritePlugin()

Method Summary

Modifier and Type	Method and Description
LatticeNode	concatenate(List<LatticeNode> path, int begin, int end, Lattice lattice, String normalizedForm) Concatenate the sequence of nodes in the path.
LatticeNode	concatenateOov(List<LatticeNode> path, int begin, int end, short posId, Lattice lattice) Concatenate the sequence of nodes in the path.
Set<CategoryType>	getCharCategoryTypes(InputText text, LatticeNode node) Return the set of the category types of the node.
abstract void	rewrite(InputText text, List<LatticeNode> path, Lattice lattice) Rewrite the path of the lattice.
void	setUp(Grammar grammar) Set up the plugin.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

settings

protected Settings settings

Constructor Detail

PathRewritePlugin

public PathRewritePlugin()

Method Detail

setUp

public void setUp(Grammar grammar)
           throws IOException

Set up the plugin. Tokenizer calls this method for setting up this plugin.

Parameters:

grammar - the grammar of the system dictionary

Throws:

IOException - if reading something is failed

rewrite

public abstract void rewrite(InputText text,
                             List<LatticeNode> path,
                             Lattice lattice)

Rewrite the path of the lattice. The path is a list of nodes of the lattice. To join some nodes you can use concatenate(java.util.List<com.worksap.nlp.sudachi.LatticeNode>, int, int, com.worksap.nlp.sudachi.Lattice, java.lang.String) or concatenateOov(java.util.List<com.worksap.nlp.sudachi.LatticeNode>, int, int, short, com.worksap.nlp.sudachi.Lattice).

Parameters:

text - the input text

path - the best path of the lattice

lattice - the lattice

concatenate

public LatticeNode concatenate(List<LatticeNode> path,
                               int begin,
                               int end,
                               Lattice lattice,
                               String normalizedForm)

Concatenate the sequence of nodes in the path. The sequence begins at the specified begin and extends to the node at index end - 1.

The concatenated node has the POS ID of the head of the sequence.

Parameters:

path - the path

begin - the beginning index

end - the ending index

lattice - the lattice

normalizedForm - if normalizedForm is null, concatenate the normalizedForms of each words

Returns:

the concatenated node

Throws:

IndexOutOfBoundsException - if begin or end are negative, greater than the length of the sequence, or begin equals or is greater than end

concatenateOov

public LatticeNode concatenateOov(List<LatticeNode> path,
                                  int begin,
                                  int end,
                                  short posId,
                                  Lattice lattice)

Concatenate the sequence of nodes in the path. The sequence begins at the specified begin and extends to the node at index end - 1.

The concatenated node is marked as OOV.

Parameters:

path - the path

begin - the beginning index

end - the ending index

posId - the POS ID of the concatenated node

lattice - the lattice

Returns:

the concatenated OOV node

Throws:

IndexOutOfBoundsException - if begin or end are negative, greater than the length of the sequence, or begin equals or is greater than end

getCharCategoryTypes

public Set<CategoryType> getCharCategoryTypes(InputText text,
                                              LatticeNode node)

Return the set of the category types of the node.

Parameters:

text - the input text

node - the node

Returns:

the set of the category types of the node