edu.washington.cs.knowitall.rule.learn
Class RuleLearner<E extends Comparable<E>>

java.lang.Object
  extended by edu.washington.cs.knowitall.rule.learn.Learner<E>
      extended by edu.washington.cs.knowitall.rule.learn.RuleLearner<E>
Direct Known Subclasses:
BeamSearchRuleLearner, MostGeneralRuleLearner

public abstract class RuleLearner<E extends Comparable<E>>
extends Learner<E>

Rule learning framework for mapping extractions onto an ontology. Annotations of rule instances are supplied, as well as corresponding extractions. For each annotation, a base rule is created by making a maximally constrained rule that results in the ontological relation given the extraction. Then the base rule is generalized using a search algorithm, by default beam search. A list of generalized rules are returned. There may be more than one value in the case of a tie.

Author:
schmmd

Field Summary
 
Fields inherited from class edu.washington.cs.knowitall.rule.learn.Learner
baseRuleFactory, confidenceFunction, validator
 
Constructor Summary
RuleLearner(BaseRuleFactory factory, RuleValidator validator, ConfidenceFunction<E> confidence)
           
 
Method Summary
 List<Rank<E>> filterBest(Iterable<Rank<E>> rules)
           
 List<Rank<E>> filterBestRules(Collection<Rule> rules)
           
 List<Rank<E>> learn(Annotation annotation)
          Generalize rules from the base rule created from the supplied annotation.
 List<Rank<E>> learn(Iterable<Annotation> annotations)
           
static void main(String[] args)
           
static void printMissedRelations(List<Rule> rules, ConfidenceFunction<?> confidence)
           
static void printRuleDerivation(Rule rule)
           
static void printRuleDerivation(Rule rule, ConfidenceFunction<?> conf)
           
static void printRuleExtractions(Rule rule, ConfidenceFunction<?> confidence)
           
abstract  List<Rank<E>> search(Rule rule)
           
 
Methods inherited from class edu.washington.cs.knowitall.rule.learn.Learner
generalize
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Constructor Detail

RuleLearner

public RuleLearner(BaseRuleFactory factory,
                   RuleValidator validator,
                   ConfidenceFunction<E> confidence)
Method Detail

learn

public List<Rank<E>> learn(Iterable<Annotation> annotations)

learn

public List<Rank<E>> learn(Annotation annotation)
Generalize rules from the base rule created from the supplied annotation.

Parameters:
annotation - from which to create the base rule.
Returns:

search

public abstract List<Rank<E>> search(Rule rule)

filterBest

public List<Rank<E>> filterBest(Iterable<Rank<E>> rules)

filterBestRules

public List<Rank<E>> filterBestRules(Collection<Rule> rules)

printRuleExtractions

public static void printRuleExtractions(Rule rule,
                                        ConfidenceFunction<?> confidence)

printMissedRelations

public static void printMissedRelations(List<Rule> rules,
                                        ConfidenceFunction<?> confidence)

printRuleDerivation

public static void printRuleDerivation(Rule rule)

printRuleDerivation

public static void printRuleDerivation(Rule rule,
                                       ConfidenceFunction<?> conf)

main

public static void main(String[] args)
                 throws org.jdom.JDOMException,
                        IOException
Throws:
org.jdom.JDOMException
IOException


Copyright © 2011 University of Washington CSE. All Rights Reserved.