Skip to content

Latest commit

 

History

History
 
 

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 
 
 
 
 
 
 

README.md

Suspecting Paraphraser 🦎 + ⌨️ → 🐍

This paraphraser transforms a yes/no question into a tag question, which helps to add more question specific informality to the dataset.

Example: "Did the American National Shipment company really break its own fleet?" -> "The American National Shipment company really broke its own fleet, didn't it?"

Author name: Witold Wydmański Author email: witold.wydmanski@uj.edu.pl Author Affiliation: Jagiellonian University

What type of a transformation is this?

This transformation acts like a paraphraser and makes lexical substitutions.

What tasks does it intend to benefit?

This transformation would benefit tasks that are focused on conversational question answering.

What are the limitations of this transformation?

The pronouns used in the endings aren't always on point - mostly because of ambiguity. As such, this transformation shouldn't be used for tasks that require their correctness.

Examples:

  • Did he drink that tea? -> He drank that tea, didn't he?
    • The pronoun gets recognized correctly
  • Has Tesla finished high school? -> Tesla has finished high school, didn't it?
    • The word Tesla gets recognized as an organization entity, thus assigned the 'it' pronoun
  • Was Sally here yesterday? -> Sally was here yesterday, wasn't [she/he/it]? (
    • Chosen at random, because any pronoun can match here. By default, names that are recognized as strongly corresponding to one of the pronouns will get 90% chance of selecting it, but it's configurable.