org.apache.stanbol.commons.opennlp
Class PosTypeChunker

java.lang.Object
  org.apache.stanbol.commons.opennlp.PosTypeChunker

Deprecated. replaced by STANBOL-733 (stanbol nlp processing module

public class PosTypeChunker
extends Object

Simple version of a Chunker that uses the POS tags to build chunks. It does not implement the Chunker interface because implementing methods other than the Chunker.chunkAsSpans(String[], String[]) is not feasible.

Defaults are based on the Penn Treebank tag set TODO:

• Test if POS tags are the same for different languages
• Check if it is possible to implement the Chunker interface

Author:

Rupert Westenthaler

Constructor Summary

PosTypeChunker(Set<String> buildPosTypes, Set<String> followPosTypes, double minPosProb)
Deprecated. Initialise a new PosTypeChunker for the parsed POS tag collections.

Method Summary

opennlp.tools.util.Span[]	chunkAsSpans(String[] tokens, String[] tags) Deprecated. Build the chunks based on the parsed tokens and POS tags.
opennlp.tools.util.Span[]	chunkAsSpans(String[] tokens, String[][] tags, double[][] props) Deprecated. Build the chunks based on the parsed tokens and the one or more detected POS tags alternatives for the tokens.
Set<String>	getChunkPosTypes() Deprecated. The set of POS types used to create Chunks
Set<String>	getFollowedPosTypes() Deprecated. The set of POS types followed to extend Chunks.
static PosTypeChunker	getInstance(String lang, double minPosTagProbaility) Deprecated. Creates an instance for the given language based on the configuration within the PosTagsCollectionEnum.

Methods inherited from class java.lang.Object

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

Constructor Detail

PosTypeChunker

public PosTypeChunker(Set<String> buildPosTypes,
                      Set<String> followPosTypes,
                      double minPosProb)

Deprecated.

Initialise a new PosTypeChunker for the parsed POS tag collections. This Constructor can be used if no predefined Configuration for a given language is available in the PosTagsCollectionEnum

Note that buildPosTypes are added to the followed once. Therefore the followPosTypes may or may not include some/all buildPosTypes.

Parameters:

buildPosTypes - the POS types that trigger a new Chunk (MUST NOT be null nor empty).

followPosTypes - additional POS types followed to extend Chunks (MAY BE null or empty).

Method Detail

getInstance

public static PosTypeChunker getInstance(String lang,
                                         double minPosTagProbaility)

Deprecated.

Creates an instance for the given language based on the configuration within the PosTagsCollectionEnum.

Parameters:

lang - The language

minPosTagProbaility - The minimum probability of a POS tag so that it is processed. In case of lower Probabilities POS tags are ignored and assumed to be matching.

Returns:

the instance or null if no configuration for the parsed language is present in the PosTagsCollectionEnum.

getFollowedPosTypes

public final Set<String> getFollowedPosTypes()

Deprecated.

The set of POS types followed to extend Chunks. This includes the getChunkPosTypes() values

Returns:

the followTypes

getChunkPosTypes

public final Set<String> getChunkPosTypes()

Deprecated.

The set of POS types used to create Chunks

Returns:

the buildTypes

chunkAsSpans

public opennlp.tools.util.Span[] chunkAsSpans(String[] tokens,
                                              String[] tags)

Deprecated.

Build the chunks based on the parsed tokens and POS tags.

This method is the equivalent to Chunker.chunkAsSpans(String[], String[])

Parameters:

tokens - the tokens

tags - the POS tags for the tokens

Returns:

the chunks as spans over the parsed tokens

chunkAsSpans

public opennlp.tools.util.Span[] chunkAsSpans(String[] tokens,
                                              String[][] tags,
                                              double[][] props)

Deprecated.

Build the chunks based on the parsed tokens and the one or more detected POS tags alternatives for the tokens.

Parameters:

tokens - the tokens

tags - the POS tags for the tokens (1D:tokens; 2D:POS tags)

Returns:

the chunks as spans over the parsed tokens