Competency Problems: On Finding and Removing Artifacts in Language Data

TL;DR

This paper introduces the concept of competency problems in NLP, emphasizing that simple feature correlations are often spurious and proposing methods to identify and mitigate dataset artifacts and biases.

Contribution

It formalizes competency problems, provides a statistical test for dataset artifacts, and analyzes the difficulty of creating unbiased datasets for complex language understanding.

Findings

Models are affected by subtle dataset biases

Realistic datasets deviate from true competency as they grow

Proposed statistical test detects spurious correlations

Abstract

Much recent work in NLP has documented dataset artifacts, bias, and spurious correlations between input features and output labels. However, how to tell which features have "spurious" instead of legitimate correlations is typically left unspecified. In this work we argue that for complex language understanding tasks, all simple feature correlations are spurious, and we formalize this notion into a class of problems which we call competency problems. For example, the word "amazing" on its own should not give information about a sentiment label independent of the context in which it appears, which could include negation, metaphor, sarcasm, etc. We theoretically analyze the difficulty of creating data for competency problems when human bias is taken into account, showing that realistic datasets will increasingly deviate from competency problems as dataset size increases. This analysis gives us a simple statistical test for dataset artifacts, which we use to show more subtle biases than were described in prior work, including demonstrating that models are inappropriately affected by these less extreme biases. Our theoretical treatment of this problem also allows us to analyze proposed solutions, such as making local edits to dataset instances, and to give recommendations for future data collection and model design efforts that target competency problems.