epic.parser.models.StructModel.scala Maven / Gradle / Ivy

Go to download
Show more of this group Show more artifacts with this name
Show all versions of epic_2.10 Show documentation
epic
The newest version!
package epic.parser
package models

/*
 Copyright 2012 David Hall

 Licensed under the Apache License, Version 2.0 (the "License")
 you may not use this file except in compliance with the License.
 You may obtain a copy of the License at

 http://www.apache.org/licenses/LICENSE-2.0

 Unless required by applicable law or agreed to in writing, software
 distributed under the License is distributed on an "AS IS" BASIS,
 WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 See the License for the specific language governing permissions and
 limitations under the License.
*/
import epic.parser.projections.GrammarRefinements
import epic.framework.Feature
import java.io.{PrintStream, FileOutputStream, File}
import breeze.linalg._
import epic.trees._
import epic.trees.annotations.TreeAnnotator
import breeze.config.Help
import epic.features._
import epic.lexicon.Lexicon
import epic.util.CacheBroker
import epic.trees.BinaryRule
import epic.trees.UnaryRule
import epic.trees.TreeInstance
import epic.trees.annotations.KMAnnotator
import epic.constraints.ChartConstraints
import epic.constraints.ChartConstraints.Factory

/**
 * Model for structural annotations, a la Klein and Manning 2003.
 * @param indexedFeatures
 * @param annotator
 * @param projections
 * @param baseFactory
 * @param topology
 * @param lexicon
 * @param initialFeatureVal
 * @tparam L
 * @tparam L2
 * @tparam W
 */
@SerialVersionUID(1L)
class StructModel[L, L2, W](indexedFeatures: IndexedFeaturizer[L, L2, W],
                        annotator: TreeAnnotator[L, W, L2],
                        val projections: GrammarRefinements[L, L2],
                        val constrainer: ChartConstraints.Factory[L, W],
                        val topology: RuleTopology[L],
                        val lexicon: Lexicon[L, W],
                        initialFeatureVal: (Feature => Option[Double]) = { _ => None }) extends ParserModel[L, W] with Serializable {
  type Inference = AnnotatedParserInference[L, W]


  def featureIndex = indexedFeatures.index

  override def initialValueForFeature(f: Feature) = initialFeatureVal(f) getOrElse 0.0

  def inferenceFromWeights(weights: DenseVector[Double]) = {
    val lexicon = new FeaturizedLexicon(weights, indexedFeatures)
    val grammar = FeaturizedGrammar(this.topology, this.lexicon, projections, weights, indexedFeatures, lexicon)
    def reannotate(tree: BinarizedTree[L], words: Seq[W]) = {
      annotator(tree, words).map(projections.labels.localize)
    }

    new AnnotatedParserInference(indexedFeatures, reannotate, grammar, constrainer)
  }


  def accumulateCounts(inf: Inference, s: Scorer, d: TreeInstance[L, W], m: Marginal, accum: ExpectedCounts, scale: Double): Unit = {
    m.expectedCounts(indexedFeatures, accum, scale)
  }
}

case class StructModelFactory(@Help(text= "The kind of annotation to do on the refined grammar. Defaults to ~KM2003")
                              annotator: TreeAnnotator[AnnotatedLabel, String, AnnotatedLabel] = KMAnnotator(),
                              @Help(text="Old weights to initialize with. Optional")
                              oldWeights: File = null,
                              annotatedTreesDumpPath: File = null) extends ParserModelFactory[AnnotatedLabel, String] {
  type MyModel = StructModel[AnnotatedLabel, AnnotatedLabel, String]


  override def make(trainTrees: IndexedSeq[TreeInstance[AnnotatedLabel, String]],
                    topology: RuleTopology[AnnotatedLabel],
                    lexicon: Lexicon[AnnotatedLabel, String],
                    constrainer: Factory[AnnotatedLabel, String]): MyModel = {
    val transformed = trainTrees.par.map(annotator).seq.toIndexedSeq

    if(annotatedTreesDumpPath != null) {
      val ps = new PrintStream(new FileOutputStream(annotatedTreesDumpPath))
      for( (x,y) <- trainTrees zip transformed) {
        ps.println("Treebank:\n" + x.render() + "\nAnnotated:\n" + y.render() + "\n==========\n")
      }
    }

    val (initLexicon, initBinaries, initUnaries) = this.extractBasicCounts(transformed)

    val refGrammar = RuleTopology(AnnotatedLabel.TOP, initBinaries, initUnaries)
    val indexedRefinements = GrammarRefinements(topology, refGrammar, (_: AnnotatedLabel).baseAnnotatedLabel)

    val surfaceFeaturizer = {
      val dsl = new WordFeaturizer.DSL(initLexicon)
      import dsl._

      (
        unigrams(word + clss, 1)
          + suffixes()
          + prefixes()
          + props
        )
    }
    val wordFeaturizer = IndexedWordFeaturizer.fromData(surfaceFeaturizer, transformed.map{_.words})
    def labelFlattener(l: AnnotatedLabel): Seq[AnnotatedLabel] = {
      val basic = Seq(l, l.baseAnnotatedLabel, l.clearFeatures).toSet.toIndexedSeq
      basic
    }

    def selectOneFeature(r: Rule[AnnotatedLabel]):IndexedSeq[Rule[AnnotatedLabel]] = {
      r match {
        case r@BinaryRule(a,b,c) =>
          val base = r.map(_.baseAnnotatedLabel)
          (( for(af <- a.features.iterator) yield base.copy(parent = base.parent.annotate(af))) ++
            ( for(af <- b.features.iterator) yield base.copy(left = base.left.annotate(af))) ++
            ( for(af <- c.features.iterator) yield base.copy(right = base.right.annotate(af)))).toIndexedSeq
        case r@UnaryRule(a,b,chain) =>
          val base = r.map(_.baseAnnotatedLabel)
          (( for(af <- a.features.iterator) yield base.copy(parent = base.parent.annotate(af))) ++
            ( for(af <- b.features.iterator) yield base.copy(child = base.child.annotate(af)))).toIndexedSeq
      }
    }

    def ruleFlattener(r: Rule[AnnotatedLabel]) = {
      (IndexedSeq(r, r.map(_.clearFeatures), r.map(_.baseAnnotatedLabel)) ++ selectOneFeature(r)).toSet.toIndexedSeq
    }
    val feat = new ProductionFeaturizer[AnnotatedLabel, AnnotatedLabel, String](topology, indexedRefinements, labelFlattener, ruleFlattener)

    val featureCounter = readWeights(oldWeights)
    val indexedFeaturizer = IndexedFeaturizer(feat, wordFeaturizer, trainTrees, annotator andThen (_.tree.map(IndexedSeq(_))), indexedRefinements)

    new StructModel[AnnotatedLabel, AnnotatedLabel, String](indexedFeaturizer,
      annotator,
      indexedRefinements,
      constrainer,
      topology,
      lexicon,
      featureCounter.get)
  }

}