Packages

class UniversalSentenceEncoder extends AnnotatorModel[UniversalSentenceEncoder] with HasSimpleAnnotate[UniversalSentenceEncoder] with HasEmbeddingsProperties with HasStorageRef with WriteTensorflowModel

The Universal Sentence Encoder encodes text into high dimensional vectors that can be used for text classification, semantic similarity, clustering and other natural language tasks.

Pretrained models can be loaded with pretrained of the companion object:

val useEmbeddings = UniversalSentenceEncoder.pretrained()
  .setInputCols("sentence")
  .setOutputCol("sentence_embeddings")

The default model is "tfhub_use", if no name is provided. For available pretrained models please see the Models Hub.

For extended examples of usage, see the Spark NLP Workshop and the UniversalSentenceEncoderTestSpec.

Sources:

Universal Sentence Encoder

https://tfhub.dev/google/universal-sentence-encoder/2

Paper abstract:

We present models for encoding sentences into embedding vectors that specifically target transfer learning to other NLP tasks. The models are efficient and result in accurate performance on diverse transfer tasks. Two variants of the encoding models allow for trade-offs between accuracy and compute resources. For both variants, we investigate and report the relationship between model complexity, resource consumption, the availability of transfer task training data, and task performance. Comparisons are made with baselines that use word level transfer learning via pretrained word embeddings as well as baselines do not use any transfer learning. We find that transfer learning using sentence embeddings tends to outperform word level transfer. With transfer learning via sentence embeddings, we observe surprisingly good performance with minimal amounts of supervised training data for a transfer task. We obtain encouraging results on Word Embedding Association Tests (WEAT) targeted at detecting model bias. Our pre-trained sentence encoding models are made freely available for download and on TF Hub.

Example

import spark.implicits._
import com.johnsnowlabs.nlp.base.DocumentAssembler
import com.johnsnowlabs.nlp.annotator.SentenceDetector
import com.johnsnowlabs.nlp.embeddings.UniversalSentenceEncoder
import com.johnsnowlabs.nlp.EmbeddingsFinisher
import org.apache.spark.ml.Pipeline

val documentAssembler = new DocumentAssembler()
  .setInputCol("text")
  .setOutputCol("document")

val sentence = new SentenceDetector()
  .setInputCols("document")
  .setOutputCol("sentence")

val embeddings = UniversalSentenceEncoder.pretrained()
  .setInputCols("sentence")
  .setOutputCol("sentence_embeddings")

val embeddingsFinisher = new EmbeddingsFinisher()
  .setInputCols("sentence_embeddings")
  .setOutputCols("finished_embeddings")
  .setOutputAsVector(true)
  .setCleanAnnotations(false)

val pipeline = new Pipeline()
  .setStages(Array(
    documentAssembler,
    sentence,
    embeddings,
    embeddingsFinisher
  ))

val data = Seq("This is a sentence.").toDF("text")
val result = pipeline.fit(data).transform(data)

result.selectExpr("explode(finished_embeddings) as result").show(5, 80)
+--------------------------------------------------------------------------------+
|                                                                          result|
+--------------------------------------------------------------------------------+
|[0.04616805538535118,0.022307956591248512,-0.044395286589860916,-0.0016493503...|
+--------------------------------------------------------------------------------+
See also

Annotators Main Page for a list of transformer based embeddings

Ordering
  1. Grouped
  2. Alphabetic
  3. By Inheritance
Inherited
  1. UniversalSentenceEncoder
  2. WriteTensorflowModel
  3. HasStorageRef
  4. HasEmbeddingsProperties
  5. HasSimpleAnnotate
  6. AnnotatorModel
  7. CanBeLazy
  8. RawAnnotator
  9. HasOutputAnnotationCol
  10. HasInputAnnotationCols
  11. HasOutputAnnotatorType
  12. ParamsAndFeaturesWritable
  13. HasFeatures
  14. DefaultParamsWritable
  15. MLWritable
  16. Model
  17. Transformer
  18. PipelineStage
  19. Logging
  20. Params
  21. Serializable
  22. Serializable
  23. Identifiable
  24. AnyRef
  25. Any
  1. Hide All
  2. Show All
Visibility
  1. Public
  2. All

Instance Constructors

  1. new UniversalSentenceEncoder()

    Annotator reference id.

    Annotator reference id. Used to identify elements in metadata or to refer to this annotator type

  2. new UniversalSentenceEncoder(uid: String)

Type Members

  1. type AnnotationContent = Seq[Row]

    internal types to show Rows as a relevant StructType Should be deleted once Spark releases UserDefinedTypes to @developerAPI

    internal types to show Rows as a relevant StructType Should be deleted once Spark releases UserDefinedTypes to @developerAPI

    Attributes
    protected
    Definition Classes
    AnnotatorModel
  2. type AnnotatorType = String
    Definition Classes
    HasOutputAnnotatorType

Value Members

  1. final def !=(arg0: Any): Boolean
    Definition Classes
    AnyRef → Any
  2. final def ##(): Int
    Definition Classes
    AnyRef → Any
  3. final def $[T](param: Param[T]): T
    Attributes
    protected
    Definition Classes
    Params
  4. def $$[T](feature: StructFeature[T]): T
    Attributes
    protected
    Definition Classes
    HasFeatures
  5. def $$[K, V](feature: MapFeature[K, V]): Map[K, V]
    Attributes
    protected
    Definition Classes
    HasFeatures
  6. def $$[T](feature: SetFeature[T]): Set[T]
    Attributes
    protected
    Definition Classes
    HasFeatures
  7. def $$[T](feature: ArrayFeature[T]): Array[T]
    Attributes
    protected
    Definition Classes
    HasFeatures
  8. final def ==(arg0: Any): Boolean
    Definition Classes
    AnyRef → Any
  9. def _transform(dataset: Dataset[_], recursivePipeline: Option[PipelineModel]): DataFrame
    Attributes
    protected
    Definition Classes
    AnnotatorModel
  10. def afterAnnotate(dataset: DataFrame): DataFrame
    Attributes
    protected
    Definition Classes
    UniversalSentenceEncoderAnnotatorModel
  11. def annotate(annotations: Seq[Annotation]): Seq[Annotation]

    Takes a document and annotations and produces new annotations of this annotator's annotation type

    Takes a document and annotations and produces new annotations of this annotator's annotation type

    annotations

    Annotations that correspond to inputAnnotationCols generated by previous annotators if any

    returns

    any number of annotations processed for every input annotation. Not necessary one to one relationship

    Definition Classes
    UniversalSentenceEncoderHasSimpleAnnotate
  12. final def asInstanceOf[T0]: T0
    Definition Classes
    Any
  13. def beforeAnnotate(dataset: Dataset[_]): Dataset[_]
    Attributes
    protected
    Definition Classes
    AnnotatorModel
  14. final def checkSchema(schema: StructType, inputAnnotatorType: String): Boolean
    Attributes
    protected
    Definition Classes
    HasInputAnnotationCols
  15. final def clear(param: Param[_]): UniversalSentenceEncoder.this.type
    Definition Classes
    Params
  16. def clone(): AnyRef
    Attributes
    protected[lang]
    Definition Classes
    AnyRef
    Annotations
    @throws( ... ) @native()
  17. val configProtoBytes: IntArrayParam

    ConfigProto from tensorflow, serialized into byte array.

    ConfigProto from tensorflow, serialized into byte array. Get with config_proto.SerializeToString()

  18. def copy(extra: ParamMap): UniversalSentenceEncoder

    requirement for annotators copies

    requirement for annotators copies

    Definition Classes
    RawAnnotator → Model → Transformer → PipelineStage → Params
  19. def copyValues[T <: Params](to: T, extra: ParamMap): T
    Attributes
    protected
    Definition Classes
    Params
  20. def createDatabaseConnection(database: Name): RocksDBConnection
    Definition Classes
    HasStorageRef
  21. final def defaultCopy[T <: Params](extra: ParamMap): T
    Attributes
    protected
    Definition Classes
    Params
  22. def dfAnnotate: UserDefinedFunction

    Wraps annotate to happen inside SparkSQL user defined functions in order to act with org.apache.spark.sql.Column

    Wraps annotate to happen inside SparkSQL user defined functions in order to act with org.apache.spark.sql.Column

    returns

    udf function to be applied to inputCols using this annotator's annotate function as part of ML transformation

    Definition Classes
    HasSimpleAnnotate
  23. val dimension: IntParam

    Number of embedding dimensions (Default: 512)

    Number of embedding dimensions (Default: 512)

    Definition Classes
    UniversalSentenceEncoderHasEmbeddingsProperties
  24. final def eq(arg0: AnyRef): Boolean
    Definition Classes
    AnyRef
  25. def equals(arg0: Any): Boolean
    Definition Classes
    AnyRef → Any
  26. def explainParam(param: Param[_]): String
    Definition Classes
    Params
  27. def explainParams(): String
    Definition Classes
    Params
  28. def extraValidate(structType: StructType): Boolean
    Attributes
    protected
    Definition Classes
    RawAnnotator
  29. def extraValidateMsg: String

    Override for additional custom schema checks

    Override for additional custom schema checks

    Attributes
    protected
    Definition Classes
    RawAnnotator
  30. final def extractParamMap(): ParamMap
    Definition Classes
    Params
  31. final def extractParamMap(extra: ParamMap): ParamMap
    Definition Classes
    Params
  32. val features: ArrayBuffer[Feature[_, _, _]]
    Definition Classes
    HasFeatures
  33. def finalize(): Unit
    Attributes
    protected[lang]
    Definition Classes
    AnyRef
    Annotations
    @throws( classOf[java.lang.Throwable] )
  34. def get[T](feature: StructFeature[T]): Option[T]
    Attributes
    protected
    Definition Classes
    HasFeatures
  35. def get[K, V](feature: MapFeature[K, V]): Option[Map[K, V]]
    Attributes
    protected
    Definition Classes
    HasFeatures
  36. def get[T](feature: SetFeature[T]): Option[Set[T]]
    Attributes
    protected
    Definition Classes
    HasFeatures
  37. def get[T](feature: ArrayFeature[T]): Option[Array[T]]
    Attributes
    protected
    Definition Classes
    HasFeatures
  38. final def get[T](param: Param[T]): Option[T]
    Definition Classes
    Params
  39. final def getClass(): Class[_]
    Definition Classes
    AnyRef → Any
    Annotations
    @native()
  40. def getConfigProtoBytes: Option[Array[Byte]]

    ConfigProto from tensorflow, serialized into byte array.

    ConfigProto from tensorflow, serialized into byte array. Get with config_proto.SerializeToString()

  41. final def getDefault[T](param: Param[T]): Option[T]
    Definition Classes
    Params
  42. def getDimension: Int

    Definition Classes
    HasEmbeddingsProperties
  43. def getInputCols: Array[String]

    returns

    input annotations columns currently used

    Definition Classes
    HasInputAnnotationCols
  44. def getLazyAnnotator: Boolean
    Definition Classes
    CanBeLazy
  45. def getLoadSP: Boolean

    Whether to load SentencePiece ops file which is required only by multi-lingual models.

  46. def getModelIfNotSet: TensorflowUSE

  47. final def getOrDefault[T](param: Param[T]): T
    Definition Classes
    Params
  48. final def getOutputCol: String

    Gets annotation column name going to generate

    Gets annotation column name going to generate

    Definition Classes
    HasOutputAnnotationCol
  49. def getParam(paramName: String): Param[Any]
    Definition Classes
    Params
  50. def getStorageRef: String
    Definition Classes
    HasStorageRef
  51. final def hasDefault[T](param: Param[T]): Boolean
    Definition Classes
    Params
  52. def hasParam(paramName: String): Boolean
    Definition Classes
    Params
  53. def hasParent: Boolean
    Definition Classes
    Model
  54. def hashCode(): Int
    Definition Classes
    AnyRef → Any
    Annotations
    @native()
  55. def initializeLogIfNecessary(isInterpreter: Boolean, silent: Boolean): Boolean
    Attributes
    protected
    Definition Classes
    Logging
  56. def initializeLogIfNecessary(isInterpreter: Boolean): Unit
    Attributes
    protected
    Definition Classes
    Logging
  57. val inputAnnotatorTypes: Array[AnnotatorType]

    Input annotator type : DOCUMENT

    Input annotator type : DOCUMENT

    Definition Classes
    UniversalSentenceEncoderHasInputAnnotationCols
  58. final val inputCols: StringArrayParam

    columns that contain annotations necessary to run this annotator AnnotatorType is used both as input and output columns if not specified

    columns that contain annotations necessary to run this annotator AnnotatorType is used both as input and output columns if not specified

    Attributes
    protected
    Definition Classes
    HasInputAnnotationCols
  59. final def isDefined(param: Param[_]): Boolean
    Definition Classes
    Params
  60. final def isInstanceOf[T0]: Boolean
    Definition Classes
    Any
  61. final def isSet(param: Param[_]): Boolean
    Definition Classes
    Params
  62. def isTraceEnabled(): Boolean
    Attributes
    protected
    Definition Classes
    Logging
  63. val lazyAnnotator: BooleanParam
    Definition Classes
    CanBeLazy
  64. val loadSP: BooleanParam

    Whether to load SentencePiece ops file which is required only by multi-lingual models (Default: false).

    Whether to load SentencePiece ops file which is required only by multi-lingual models (Default: false). This is not changeable after it's set with a pretrained model nor it is compatible with Windows.

  65. def log: Logger
    Attributes
    protected
    Definition Classes
    Logging
  66. def logDebug(msg: ⇒ String, throwable: Throwable): Unit
    Attributes
    protected
    Definition Classes
    Logging
  67. def logDebug(msg: ⇒ String): Unit
    Attributes
    protected
    Definition Classes
    Logging
  68. def logError(msg: ⇒ String, throwable: Throwable): Unit
    Attributes
    protected
    Definition Classes
    Logging
  69. def logError(msg: ⇒ String): Unit
    Attributes
    protected
    Definition Classes
    Logging
  70. def logInfo(msg: ⇒ String, throwable: Throwable): Unit
    Attributes
    protected
    Definition Classes
    Logging
  71. def logInfo(msg: ⇒ String): Unit
    Attributes
    protected
    Definition Classes
    Logging
  72. def logName: String
    Attributes
    protected
    Definition Classes
    Logging
  73. def logTrace(msg: ⇒ String, throwable: Throwable): Unit
    Attributes
    protected
    Definition Classes
    Logging
  74. def logTrace(msg: ⇒ String): Unit
    Attributes
    protected
    Definition Classes
    Logging
  75. def logWarning(msg: ⇒ String, throwable: Throwable): Unit
    Attributes
    protected
    Definition Classes
    Logging
  76. def logWarning(msg: ⇒ String): Unit
    Attributes
    protected
    Definition Classes
    Logging
  77. def msgHelper(schema: StructType): String
    Attributes
    protected
    Definition Classes
    HasInputAnnotationCols
  78. final def ne(arg0: AnyRef): Boolean
    Definition Classes
    AnyRef
  79. final def notify(): Unit
    Definition Classes
    AnyRef
    Annotations
    @native()
  80. final def notifyAll(): Unit
    Definition Classes
    AnyRef
    Annotations
    @native()
  81. def onWrite(path: String, spark: SparkSession): Unit
  82. val optionalInputAnnotatorTypes: Array[String]
    Definition Classes
    HasInputAnnotationCols
  83. val outputAnnotatorType: AnnotatorType

    Output annotator type : SENTENCE_EMBEDDINGS

    Output annotator type : SENTENCE_EMBEDDINGS

    Definition Classes
    UniversalSentenceEncoderHasOutputAnnotatorType
  84. final val outputCol: Param[String]
    Attributes
    protected
    Definition Classes
    HasOutputAnnotationCol
  85. lazy val params: Array[Param[_]]
    Definition Classes
    Params
  86. var parent: Estimator[UniversalSentenceEncoder]
    Definition Classes
    Model
  87. def save(path: String): Unit
    Definition Classes
    MLWritable
    Annotations
    @Since( "1.6.0" ) @throws( ... )
  88. def set[T](feature: StructFeature[T], value: T): UniversalSentenceEncoder.this.type
    Attributes
    protected
    Definition Classes
    HasFeatures
  89. def set[K, V](feature: MapFeature[K, V], value: Map[K, V]): UniversalSentenceEncoder.this.type
    Attributes
    protected
    Definition Classes
    HasFeatures
  90. def set[T](feature: SetFeature[T], value: Set[T]): UniversalSentenceEncoder.this.type
    Attributes
    protected
    Definition Classes
    HasFeatures
  91. def set[T](feature: ArrayFeature[T], value: Array[T]): UniversalSentenceEncoder.this.type
    Attributes
    protected
    Definition Classes
    HasFeatures
  92. final def set(paramPair: ParamPair[_]): UniversalSentenceEncoder.this.type
    Attributes
    protected
    Definition Classes
    Params
  93. final def set(param: String, value: Any): UniversalSentenceEncoder.this.type
    Attributes
    protected
    Definition Classes
    Params
  94. final def set[T](param: Param[T], value: T): UniversalSentenceEncoder.this.type
    Definition Classes
    Params
  95. def setConfigProtoBytes(bytes: Array[Int]): UniversalSentenceEncoder.this.type

    ConfigProto from tensorflow, serialized into byte array.

    ConfigProto from tensorflow, serialized into byte array. Get with config_proto.SerializeToString()

  96. def setDefault[T](feature: StructFeature[T], value: () ⇒ T): UniversalSentenceEncoder.this.type
    Attributes
    protected
    Definition Classes
    HasFeatures
  97. def setDefault[K, V](feature: MapFeature[K, V], value: () ⇒ Map[K, V]): UniversalSentenceEncoder.this.type
    Attributes
    protected
    Definition Classes
    HasFeatures
  98. def setDefault[T](feature: SetFeature[T], value: () ⇒ Set[T]): UniversalSentenceEncoder.this.type
    Attributes
    protected
    Definition Classes
    HasFeatures
  99. def setDefault[T](feature: ArrayFeature[T], value: () ⇒ Array[T]): UniversalSentenceEncoder.this.type
    Attributes
    protected
    Definition Classes
    HasFeatures
  100. final def setDefault(paramPairs: ParamPair[_]*): UniversalSentenceEncoder.this.type
    Attributes
    protected
    Definition Classes
    Params
  101. final def setDefault[T](param: Param[T], value: T): UniversalSentenceEncoder.this.type
    Attributes
    protected
    Definition Classes
    Params
  102. def setDimension(value: Int): UniversalSentenceEncoder.this.type

    Definition Classes
    HasEmbeddingsProperties
  103. final def setInputCols(value: String*): UniversalSentenceEncoder.this.type
    Definition Classes
    HasInputAnnotationCols
  104. def setInputCols(value: Array[String]): UniversalSentenceEncoder.this.type

    Overrides required annotators column if different than default

    Overrides required annotators column if different than default

    Definition Classes
    HasInputAnnotationCols
  105. def setLazyAnnotator(value: Boolean): UniversalSentenceEncoder.this.type
    Definition Classes
    CanBeLazy
  106. def setLoadSP(value: Boolean): UniversalSentenceEncoder.this.type

    Whether to load SentencePiece ops file which is required only by multi-lingual models.

  107. def setModelIfNotSet(spark: SparkSession, tensorflow: TensorflowWrapper): UniversalSentenceEncoder.this.type

  108. final def setOutputCol(value: String): UniversalSentenceEncoder.this.type

    Overrides annotation column name when transforming

    Overrides annotation column name when transforming

    Definition Classes
    HasOutputAnnotationCol
  109. def setParent(parent: Estimator[UniversalSentenceEncoder]): UniversalSentenceEncoder
    Definition Classes
    Model
  110. def setStorageRef(value: String): UniversalSentenceEncoder.this.type
    Definition Classes
    HasStorageRef
  111. val storageRef: Param[String]

    Unique identifier for storage (Default: this.uid)

    Unique identifier for storage (Default: this.uid)

    Definition Classes
    HasStorageRef
  112. final def synchronized[T0](arg0: ⇒ T0): T0
    Definition Classes
    AnyRef
  113. def toString(): String
    Definition Classes
    Identifiable → AnyRef → Any
  114. final def transform(dataset: Dataset[_]): DataFrame

    Given requirements are met, this applies ML transformation within a Pipeline or stand-alone Output annotation will be generated as a new column, previous annotations are still available separately metadata is built at schema level to record annotations structural information outside its content

    Given requirements are met, this applies ML transformation within a Pipeline or stand-alone Output annotation will be generated as a new column, previous annotations are still available separately metadata is built at schema level to record annotations structural information outside its content

    dataset

    Dataset[Row]

    Definition Classes
    AnnotatorModel → Transformer
  115. def transform(dataset: Dataset[_], paramMap: ParamMap): DataFrame
    Definition Classes
    Transformer
    Annotations
    @Since( "2.0.0" )
  116. def transform(dataset: Dataset[_], firstParamPair: ParamPair[_], otherParamPairs: ParamPair[_]*): DataFrame
    Definition Classes
    Transformer
    Annotations
    @Since( "2.0.0" ) @varargs()
  117. final def transformSchema(schema: StructType): StructType

    requirement for pipeline transformation validation.

    requirement for pipeline transformation validation. It is called on fit()

    Definition Classes
    RawAnnotator → PipelineStage
  118. def transformSchema(schema: StructType, logging: Boolean): StructType
    Attributes
    protected
    Definition Classes
    PipelineStage
    Annotations
    @DeveloperApi()
  119. val uid: String
    Definition Classes
    UniversalSentenceEncoder → Identifiable
  120. def validate(schema: StructType): Boolean

    takes a Dataset and checks to see if all the required annotation types are present.

    takes a Dataset and checks to see if all the required annotation types are present.

    schema

    to be validated

    returns

    True if all the required types are present, else false

    Attributes
    protected
    Definition Classes
    RawAnnotator
  121. def validateStorageRef(dataset: Dataset[_], inputCols: Array[String], annotatorType: String): Unit
    Definition Classes
    HasStorageRef
  122. final def wait(): Unit
    Definition Classes
    AnyRef
    Annotations
    @throws( ... )
  123. final def wait(arg0: Long, arg1: Int): Unit
    Definition Classes
    AnyRef
    Annotations
    @throws( ... )
  124. final def wait(arg0: Long): Unit
    Definition Classes
    AnyRef
    Annotations
    @throws( ... ) @native()
  125. def wrapColumnMetadata(col: Column): Column
    Attributes
    protected
    Definition Classes
    RawAnnotator
  126. def wrapEmbeddingsMetadata(col: Column, embeddingsDim: Int, embeddingsRef: Option[String] = None): Column
    Attributes
    protected
    Definition Classes
    HasEmbeddingsProperties
  127. def wrapSentenceEmbeddingsMetadata(col: Column, embeddingsDim: Int, embeddingsRef: Option[String] = None): Column
    Attributes
    protected
    Definition Classes
    HasEmbeddingsProperties
  128. def write: MLWriter
    Definition Classes
    ParamsAndFeaturesWritable → DefaultParamsWritable → MLWritable
  129. def writeTensorflowHub(path: String, tfPath: String, spark: SparkSession, suffix: String = "_use"): Unit
    Definition Classes
    WriteTensorflowModel
  130. def writeTensorflowModel(path: String, spark: SparkSession, tensorflow: TensorflowWrapper, suffix: String, filename: String, configProtoBytes: Option[Array[Byte]] = None): Unit
    Definition Classes
    WriteTensorflowModel
  131. def writeTensorflowModelV2(path: String, spark: SparkSession, tensorflow: TensorflowWrapper, suffix: String, filename: String, configProtoBytes: Option[Array[Byte]] = None, savedSignatures: Option[Map[String, String]] = None): Unit
    Definition Classes
    WriteTensorflowModel

Inherited from WriteTensorflowModel

Inherited from HasStorageRef

Inherited from HasEmbeddingsProperties

Inherited from CanBeLazy

Inherited from HasOutputAnnotationCol

Inherited from HasInputAnnotationCols

Inherited from HasOutputAnnotatorType

Inherited from ParamsAndFeaturesWritable

Inherited from HasFeatures

Inherited from DefaultParamsWritable

Inherited from MLWritable

Inherited from Model[UniversalSentenceEncoder]

Inherited from Transformer

Inherited from PipelineStage

Inherited from Logging

Inherited from Params

Inherited from Serializable

Inherited from Serializable

Inherited from Identifiable

Inherited from AnyRef

Inherited from Any

Parameters

A list of (hyper-)parameter keys this annotator can take. Users can set and get the parameter values through setters and getters, respectively.

Annotator types

Required input and expected output annotator types

Members

Parameter setters

Parameter getters