Measuring Attribution in Natural Language Generation Models

Authors:
  • HRHannah Rashkin
    Hannah Rashkin
  • VNVitaly Nikolaev
    Vitaly Nikolaev
  • MLMatthew Lamm
    Matthew Lamm
  • MCMichael Collins
    Michael Collins
  • DDDipanjan Das
    Dipanjan Das
  • SPSlav Petrov
    Slav Petrov
  • GTGaurav Singh Tomar
    Gaurav Singh Tomar
  • ITIulia Turc
    Iulia Turc
  • DRDavid Reitter
    David Reitter

Abstract
With recent improvements in natural language generation (NLG) models for various applications, it has become imperative to have the means to identify and evaluate whether NLG output is only sharing verifiable information about the external world. In this work, we present a new evaluation framework entitled Attributable to Identified Sources (AIS) for assessing the output of natural language generation models, when such output pertains to the external world. We first define AIS and introduce a two-stage annotation pipeline for allowing annotators to appropriately evaluate model output according to AIS guidelines. We empirically validate this approach on three generation datasets (two in the conversational QA domain and one in summarization) via human evaluation studies that suggest that AIS could serve as a common framework for measuring whether model-generated statements are supported by underlying sources. We release guidelines for the human evaluation studies.