False perfection in machine prediction: Detecting and assessing circularity problems in machine learning

TL;DR

This paper discusses the issue of circularity in machine learning predictions, highlighting how it can lead to false perceptions of accuracy and proposing methods to detect and assess such problems.

Contribution

It introduces a framework for identifying and evaluating circularity issues in machine learning models, emphasizing the importance of validity and reliability.

Findings

Circularity can cause false confidence in model predictions

Detection methods can identify circularity problems

Assessment techniques improve model validity

Abstract

This paper is an excerpt of an early version of Chapter 2 of the book "Validity, Reliability, and Significance. Empirical Methods for NLP and Data Science", by Stefan Riezler and Michael Hagmann, published in December 2021 by Morgan & Claypool. Please see the book's homepage at https://www.morganclaypoolpublishers.com/catalog_Orig/product_info.php?products_id=1688 for a more recent and comprehensive discussion.