ml

Type Members

1. case class BrownClusters(clusters: Iterable[(Symbol, Seq[Int])]) extends Product with Serializable

   

2. case class BrownClustersTagger(clusters: Seq[BrownClusters]) extends TokenTagger with Product with Serializable

   

   The BrownClustersTagger tags the tokens of a sentence with their Brown clusters.

3. case class DatastoreGoogleNGram(groupName: String, artifactName: String, version: Int, frequencyCutoff: Int) extends Product with Serializable

   

   format: OFF A class that parses Google N-Gram data (http://commondatastorage.googleapis.com/books/syntactic-ngrams/index.html) to provide information about a requested n-gram.

   format: OFF A class that parses Google N-Gram data (http://commondatastorage.googleapis.com/books/syntactic-ngrams/index.html) to provide information about a requested n-gram. Takes the datastore location details for a data directory and parses each file, expected to be in the following format (from https://docs.google.com/document/d/14PWeoTkrnKk9H8_7CfVbdvuoFZ7jYivNTkBX2Hj7qLw/edit) - head_word<TAB>syntactic-ngram<TAB>total_count<TAB>counts_by_year The counts_by_year format is a tab-separated list of year<comma>count items. Years are sorted in ascending order, and only years with non-zero counts are included. The syntactic-ngram format is a space-separated list of tokens, each token format is: “word/pos-tag/dep-label/head-index”. The word field can contain any non-whitespace character. The other fields can contain any non-whitespace character except for ‘/’. pos-tag is a Penn-Treebank part-of-speech tag. dep-label is a stanford-basic-dependencies label. head-index is an integer, pointing to the head of the current token. “1” refers to the first token in the list, 2 the second, and 0 indicates that the head is the root of the fragment. format: ON

4. case class FeatureEncoding(featureNames: IndexedSeq[FeatureName]) extends Product with Serializable

   

   Maps feature names to integers.

   Maps feature names to integers. Useful for serializing TrainingData instances for consumption by command-line machine learning tools.

   featureNames

   an indexed sequence of feature names

5. case class FeatureName(symbols: Seq[Symbol]) extends Product with Serializable

   

   The name of a feature, represented as a list of Symbols.

   The name of a feature, represented as a list of Symbols.

   symbols

   the list of symbols comprising the feature name

6. case class FeatureVector(values: Seq[(FeatureName, Double)]) extends Product with Serializable

   

   A mapping from feature names to values.

   A mapping from feature names to values.

   Unspecified feature names are assumed to correspond to a value of zero.

   values

   the map from feature names to values

7. sealed trait GoogleUnigramTagType extends AnyRef

   

8. case class GoogleUnigramTagger(googleNgram: DatastoreGoogleNGram, tagType: GoogleUnigramTagType) extends TokenTagger with Product with Serializable

   

   A SentenceTagger that tags tokens with unigram features from Google N-grams.

   A SentenceTagger that tags tokens with unigram features from Google N-grams.

   googleNgram

   the Google N-grams resource

   tagType

   the Google N-grams tag type you want to create features for

9. case class LinearModel(coefficients: Seq[(FeatureName, Double)]) extends Product with Serializable

   

   A weighted linear combination of features.

   A weighted linear combination of features.

   coefficients

   map from feature names to weight coefficients

10. case class NgramInfo(syntacticNgram: Seq[SyntacticInfo], frequency: Long) extends Product with Serializable

    

11. case class SyntacticInfo(word: String, posTag: String, depLabel: String, headIndex: Int) extends Product with Serializable

    

    Utility case classes to represent information associated with an ngram in the Google Ngram corpus.

12. case class TrainingData(labeledVectors: Iterable[(FeatureVector, Int)]) extends Product with Serializable

    

    Abstraction for a set of labeled feature vectors.

    Abstraction for a set of labeled feature vectors.

    Provides various serialization options for different machine learning tools.

    labeledVectors

    a sequence of feature vectors labeled with integer outcomes

13. case class UnigramInfo(syntacticUnigram: SyntacticInfo, frequency: Long) extends Product with Serializable

    

    Encapsulates unigram info pertaining to a word.

    Encapsulates unigram info pertaining to a word. Instead of a seq of SyntacticInfo objects in the general purpose NgramInfo class, here we have a single SyntacticInfo representing the info for a single gram.

14. case class Verbnet(groupName: String, artifactName: String, version: Int) extends Product with Serializable

    

    A class that uses JVerbnet, a 3rd party Wrapper library for Verbnet data (http://projects.csail.mit.edu/jverbnet/), to quickly look up various verbnet features for a verb.

15. case class VerbnetTagger(verbnet: Verbnet, useSecondaryFrames: Boolean = false) extends SentenceTagger with Product with Serializable

    

    A SentenceTagger that tags sentence tokens using Verbnet frames.

    A SentenceTagger that tags sentence tokens using Verbnet frames.

    verbnet

    the associated Verbnet resource

    useSecondaryFrames

    set to true if you want secondary (rather than primary) frames

16. case class WrapperClassifier(classifier: ProbabilisticClassifier, featureNameMap: Seq[(Int, FeatureName)]) extends Product with Serializable

    

    A WrapperClassifier wraps a ProbabilisticClassifier (which uses integer-based feature names) in an interface that allows you to use the more natural org.allenai.nlpstack.parse.poly.ml FeatureVector format for classification.

    A WrapperClassifier wraps a ProbabilisticClassifier (which uses integer-based feature names) in an interface that allows you to use the more natural org.allenai.nlpstack.parse.poly.ml FeatureVector format for classification. This is a trait that specific wrappers can extend.

17. class WrapperClassifierTrainer extends AnyRef

    

    Trains a WrapperClassifier from training data.