edu.stanford.nlp.parser.oldparser
Class ChartParser

java.lang.Object
  |
  +--edu.stanford.nlp.parser.oldparser.ChartParser

public final class ChartParser

extends Object

ChartParser. This is my haiku about this parser: Chart parser of mine, Experimental but fast, Please parse correctly. That pretty much sums it up.

Field Summary

static boolean	ALL_PARSES Keep all parses for a sentence in the chart
static boolean	BEST_ONLY Keep only best edges over a span in a chart, based on edge score
static int	BOTTOM_UP Argument to constructor: parse bottom up
static boolean	EMPTIES_OFF Do not place empty nodes in the chart
static boolean	EMPTIES_ON Place an empty (edu.stanford.nlp.trees.Word.EMPTY at each chart vertex, so that rules with empties on the RHS will parse.
static boolean	SCANNER_OFF Argument to constructor: place all words into chart on initialization
static boolean	SCANNER_ON Argument to constructor: scan (shift) words one at a time, as in the original Earley algorithm
static boolean	STATISTICS_OFF Do not record statistics on Agenda operations
static boolean	STATISTICS_ON Record some statistics on Agenda operations
static int	TOP_DOWN Argument to constructor: parse top down

Constructor Summary

ChartParser(DFSAGrammar grammar, Lexicon lexicon, int direction, boolean scanMode, boolean statistics, boolean bestOnly)
Builds a parser for a grammar and lexicon.

ChartParser(DFSAGrammar grammar, Lexicon lexicon, int direction, boolean scanMode, boolean statistics, boolean bestOnly, boolean empties)
Builds a parser for a grammar and lexicon.

Method Summary

void	analyzeChartRelaxation()
void	analyzeChartSaturation()
void	dumpChart()
List	getAllParses(Label label)
List	getAllParses(LabeledConstituent c)
Tree	getBestParse(Label label)
Tree	getBestParse(LabeledConstituent c)
long	getNumEdges()
long	getNumEdgesAtGoal()
long	getNumTraversals()
long	getNumTraversalsAtGoal()
void	parse(Sentence s, Label goal) The basic call for parsing a sentence once the chart structure has been created.

Methods inherited from class java.lang.Object

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

Field Detail

BOTTOM_UP

public static final int BOTTOM_UP

Argument to constructor: parse bottom up

See Also:

Constant Field Values

TOP_DOWN

public static final int TOP_DOWN

Argument to constructor: parse top down

See Also:

Constant Field Values

SCANNER_ON

public static final boolean SCANNER_ON

Argument to constructor: scan (shift) words one at a time, as in the original Earley algorithm

See Also:

Constant Field Values

SCANNER_OFF

public static final boolean SCANNER_OFF

Argument to constructor: place all words into chart on initialization

See Also:

Constant Field Values

EMPTIES_ON

public static final boolean EMPTIES_ON

Place an empty (edu.stanford.nlp.trees.Word.EMPTY at each chart vertex, so that rules with empties on the RHS will parse.

See Also:

Constant Field Values

EMPTIES_OFF

public static final boolean EMPTIES_OFF

Do not place empty nodes in the chart

See Also:

Constant Field Values

STATISTICS_ON

public static final boolean STATISTICS_ON

Record some statistics on Agenda operations

See Also:

Constant Field Values

STATISTICS_OFF

public static final boolean STATISTICS_OFF

Do not record statistics on Agenda operations

See Also:

Constant Field Values

ALL_PARSES

public static final boolean ALL_PARSES

Keep all parses for a sentence in the chart

See Also:

Constant Field Values

BEST_ONLY

public static final boolean BEST_ONLY

Keep only best edges over a span in a chart, based on edge score

See Also:

Constant Field Values

Constructor Detail

ChartParser

public ChartParser(DFSAGrammar grammar,
                   Lexicon lexicon,
                   int direction,
                   boolean scanMode,
                   boolean statistics,
                   boolean bestOnly)

Builds a parser for a grammar and lexicon.

ChartParser

public ChartParser(DFSAGrammar grammar,
                   Lexicon lexicon,
                   int direction,
                   boolean scanMode,
                   boolean statistics,
                   boolean bestOnly,
                   boolean empties)

Builds a parser for a grammar and lexicon.

Method Detail

getBestParse

public Tree getBestParse(Label label)

getBestParse

public Tree getBestParse(LabeledConstituent c)

getAllParses

public List getAllParses(Label label)

getAllParses

public List getAllParses(LabeledConstituent c)

getNumEdges

public long getNumEdges()

getNumTraversals

public long getNumTraversals()

getNumEdgesAtGoal

public long getNumEdgesAtGoal()

getNumTraversalsAtGoal

public long getNumTraversalsAtGoal()

parse

public void parse(Sentence s,
                  Label goal)

The basic call for parsing a sentence once the chart structure has been created. To parse from the tag level, the easiest supported thing to do is to make lexicon null, and then to pass in a sentence of type TaggedWord.

Parameters:

s - The sentence to be parsed

goal - The Label of the goal

analyzeChartSaturation

public void analyzeChartSaturation()

analyzeChartRelaxation

public void analyzeChartRelaxation()

dumpChart

public void dumpChart()