Deconstructing NLG Evaluation: Evaluation Practices, Assumptions, and Their Implications

TL;DR

This paper investigates the diverse evaluation practices of NLG practitioners, revealing underlying goals, assumptions, and ethical considerations through interviews and surveys to better understand evaluation challenges and implications.

Contribution

It provides a comprehensive analysis of practitioner goals and assumptions in NLG evaluation, highlighting implicit practices and ethical considerations often overlooked.

Findings

Practitioners have diverse evaluation goals and methods.

Evaluation assumptions often remain implicit or unstated.

Ethical considerations influence evaluation practices.

Abstract

There are many ways to express similar things in text, which makes evaluating natural language generation (NLG) systems difficult. Compounding this difficulty is the need to assess varying quality criteria depending on the deployment setting. While the landscape of NLG evaluation has been well-mapped, practitioners' goals, assumptions, and constraints -- which inform decisions about what, when, and how to evaluate -- are often partially or implicitly stated, or not stated at all. Combining a formative semi-structured interview study of NLG practitioners (N=18) with a survey study of a broader sample of practitioners (N=61), we surface goals, community practices, assumptions, and constraints that shape NLG evaluations, examining their implications and how they embody ethical considerations.