Package

epic

sequences

Permalink

package sequences

Visibility
  1. Public
  2. All

Type Members

  1. sealed trait BIOETag[+L] extends AnyRef

    Permalink

    A BIOETag is a tag that us to represent epic.sequences.Segmentations as epic.sequences.TaggedSequences.

    A BIOETag is a tag that us to represent epic.sequences.Segmentations as epic.sequences.TaggedSequences. It includes Begins, Inside, Outside, and End tags. Sometimes we just use IO, or BIO.

  2. trait CRF[L, W] extends Serializable

    Permalink

    A Linear Chain Conditional Random Field.

    A Linear Chain Conditional Random Field. Useful for POS tagging, etc.

    As usual in Epic, all the heavy lifting is done in the companion object and Marginals.

    CRFs can produce epic.sequences.TaggedSequence from an input sequence of words. They can also produce marginals, etc.

    Annotations
    @SerialVersionUID()
  3. class CRFInference[L, W] extends AugmentableInference[TaggedSequence[L, W], Anchoring[L, W]] with CRF[L, W] with AnnotatingInference[TaggedSequence[L, W]] with Serializable

    Permalink
    Annotations
    @SerialVersionUID()
  4. class CRFModel[L, W] extends Model[TaggedSequence[L, W]] with Model[TaggedSequence[L, W]] with Serializable

    Permalink

    Annotations
    @SerialVersionUID()
  5. trait Gazetteer[+L, W] extends SurfaceFeaturizer[W] with WordFeaturizer[W]

    Permalink

    A Gazeteer is a map from IndexedSeq[W]->L.

    A Gazeteer is a map from IndexedSeq[W]->L. That is, it maps strings of words to a label that we've seen before. For example, you might use a list of countries. These are very useful for named entity recognition.

  6. case class GazetteerSpanFeature(label: Any) extends Feature with Product with Serializable

    Permalink
  7. case class GazetteerWordFeature(label: Any) extends Feature with Product with Serializable

    Permalink
  8. trait GoldSegmentPolicy[L] extends AnyRef

    Permalink
  9. class HammingLossAugmentation[L, W] extends LossAugmentation[Segmentation[L, W], Anchoring[L, W]]

    Permalink

    TODO

  10. case class Segmentation[+L, +W](segments: IndexedSeq[(L, Span)], words: IndexedSeq[W], id: String = "") extends Example[IndexedSeq[(L, Span)], IndexedSeq[W]] with Product with Serializable

    Permalink

  11. class SegmentationModelFactory[L] extends SerializableLogging

    Permalink

    Factory class for making a epic.sequences.SemiCRFModel based on some data and an optional gazetteer.

  12. trait Segmenter[Tag] extends StringAnalysisFunction[Sentence with Token, Tag] with (IndexedSeq[String]) ⇒ IndexedSeq[(Tag, Span)]

    Permalink

    A epic.sequences.Segmenter splits up a sentence into labeled segments.

    A epic.sequences.Segmenter splits up a sentence into labeled segments. For instance, it might find all the people, places and things (Named Entity Recognition) in a document.

    Tag

    the type of tag that is annotated

  13. trait SemiCRF[L, W] extends Serializable

    Permalink

    A Semi-Markov Linear Chain Conditional Random Field, that is, the length of time spent in a state may be longer than 1 tick.

    A Semi-Markov Linear Chain Conditional Random Field, that is, the length of time spent in a state may be longer than 1 tick. Useful for field segmentation or NER.

    As usual in Epic, all the heavy lifting is done in the companion object and Marginals.

    Annotations
    @SerialVersionUID()
  14. class SemiCRFInference[L, W] extends AugmentableInference[Segmentation[L, W], Anchoring[L, W]] with SemiCRF[L, W] with Serializable

    Permalink
    Annotations
    @SerialVersionUID()
  15. class SemiCRFModel[L, W] extends Model[Segmentation[L, W]] with Serializable

    Permalink

    Annotations
    @SerialVersionUID()
  16. case class TaggedSequence[+L, +W](tags: IndexedSeq[L], words: IndexedSeq[W], id: String = "") extends Example[IndexedSeq[L], IndexedSeq[W]] with Product with Serializable

    Permalink

    A tagged sequence has a sequence of tags and a sequence of words that are in one-to-one correspondence.

    A tagged sequence has a sequence of tags and a sequence of words that are in one-to-one correspondence. think POS tags etc.

  17. class TaggedSequenceModelFactory[L] extends SerializableLogging

    Permalink
  18. trait Tagger[Tag] extends StringAnalysisFunction[Sentence with Token, Tag] with (IndexedSeq[String]) ⇒ IndexedSeq[Tag]

    Permalink

    A Tagger assigns a sequence of Tags to a

    A Tagger assigns a sequence of Tags to a

    Tag

    the type of tag that is annotated

Value Members

  1. object BIOETag

    Permalink
  2. object CRF extends Serializable

    Permalink
  3. object Gazetteer extends Serializable

    Permalink
  4. object GoldSegmentPolicy

    Permalink
  5. object HMM

    Permalink

    HiddenMarkovModel, which is the generative special case of a epic.sequences.CRF.

  6. object SegmentText extends ProcessTextMain[SemiCRF[Any, String], Segmentation[Any, String]]

    Permalink

    Simple class that reads in a bunch of files and parses them.

    Simple class that reads in a bunch of files and parses them. Output is dumped to standard out.

  7. object Segmentation extends Serializable

    Permalink
  8. object SegmentationEval extends SerializableLogging

    Permalink

    Object for evaluating epic.sequences.Segmentations.

    Object for evaluating epic.sequences.Segmentations. Returned metrics are precision, recall, and f1

  9. object SegmentationModelFactory extends Serializable

    Permalink
    Annotations
    @SerialVersionUID()
  10. object Segmenter

    Permalink
  11. object SemiCRF extends Serializable

    Permalink
  12. object SemiCRFModel extends Serializable

    Permalink
  13. object SemiConllNerPipeline extends SerializableLogging

    Permalink
  14. object SemiNerPipeline extends SerializableLogging

    Permalink

  15. object SemiPOSTagger extends SerializableLogging

    Permalink

    Mostly for debugging SemiCRFs.

    Mostly for debugging SemiCRFs. Just uses a SemiCRF as a CRF.

  16. object TagText extends ProcessTextMain[CRF[AnnotatedLabel, String], TaggedSequence[AnnotatedLabel, String]]

    Permalink

    Simple class that reads in a bunch of files and tags them.

    Simple class that reads in a bunch of files and tags them. Output is dumped to standard out.

  17. object TaggedSequenceEval

    Permalink

    Object for evaluating epic.sequences.TaggedSequences.

    Object for evaluating epic.sequences.TaggedSequences. Returned metrics are accuracy and exact match.

  18. object TaggedSequenceModelFactory extends Serializable

    Permalink
  19. object Tagger

    Permalink
  20. object TrainPosTagger extends SerializableLogging

    Permalink

Ungrouped