class NerDatasetEncoder extends Serializable
Linear Supertypes
Ordering
- Alphabetic
- By Inheritance
Inherited
- NerDatasetEncoder
- Serializable
- Serializable
- AnyRef
- Any
- Hide All
- Show All
Visibility
- Public
- All
Instance Constructors
- new NerDatasetEncoder(params: DatasetEncoderParams)
Value Members
- val char2Id: Map[Char, Int]
- val chars: Array[Char]
-
def
convertBatchTags(predictedTags: Array[String], allTags: Array[String], sentenceLength: Array[Int], prob: Option[Seq[Array[Float]]], includeAllConfidenceScores: Boolean): Array[Array[(String, Option[Array[Map[String, String]]])]]
Converts Tensorflow tags output to 2-dimensional Array with shape: (Batch, Sentence Length).
Converts Tensorflow tags output to 2-dimensional Array with shape: (Batch, Sentence Length).
- predictedTags
2-dimensional tensor in plain array
- allTags
All original tags
- sentenceLength
Every sentence length (number of words).
- returns
List of tags for each sentence
-
def
decodeOutputData(tagIds: Array[Int]): Array[String]
Converts Tag Identifiers to Source Names
Converts Tag Identifiers to Source Names
- tagIds
Tag Ids encoded for Tensorflow Model.
- returns
Tag names
- def encodeInputData(sentences: Array[WordpieceEmbeddingsSentence]): NerBatch
-
def
encodeTags(tags: Array[Array[String]]): Array[Array[Int]]
Converts Tag names to Identifiers
Converts Tag names to Identifiers
- tags
batches of labels/classes for each sentence/document
- returns
batches of tag ids for each sentence/document
- def getOrElse[T](source: Array[T], i: Int, value: ⇒ T): T
- val params: DatasetEncoderParams
- val tag2Id: Map[String, Int]
- val tags: Array[String]