Class NgramRelativeTokenDistance

This algorithm calculates the relative distance between tokens within an n-gram. NOTE: this algorithm is only useful for dis-contiguous n-grams.

Some words that are textually identical but have different occurrences may be falsely suggested with a word farther away within the sentence.

This algorithm can be used to correct those false positives. Results range from 0 to 1.

Hierarchy

Algorithm
- NgramRelativeTokenDistance

Implements

AlgorithmType

Index

Properties

name

Methods

Properties

name

name: string = "alignment relative occurrence"

Methods

execute

execute(prediction: Prediction, cIndex: CorpusIndex, saIndex: AlignmentMemoryIndex, usIndex: UnalignedSentenceIndex): Prediction

Overrides Algorithm.execute
- Defined in algorithms/NgramRelativeTokenDistance.ts:43
Parameters
- prediction: Prediction
- cIndex: CorpusIndex
- saIndex: AlignmentMemoryIndex
- usIndex: UnalignedSentenceIndex
Returns Prediction

Static calculate

calculate(t: number, x: number, y: number): number

- Defined in algorithms/NgramRelativeTokenDistance.ts:27
Calculates the relative distance between positions. A score of 0 means the tokens are on opposite sides of the sentence. A score of 1 means they are next to each other.

Parameters
- t: number
  
  length of sentence
- x: number
  
  first position
- y: number
  
  second position
Returns number

Generated using TypeDoc