Evaluating Machine Common Sense via Cloze Testing

TL;DR

This paper assesses the extent to which language models truly understand common sense by systematically testing their performance and confidence levels using cloze tests and word embeddings.

Contribution

It introduces a novel framework for evaluating LM's common sense through a series of tests measuring robustness and confidence, highlighting current limitations.

Findings

Language models achieve human-like accuracy on CS questions.

Models' confidence levels are generally lower than expected.

Results suggest potential for combining symbolic and distributed knowledge.

Abstract

Language models (LMs) show state of the art performance for common sense (CS) question answering, but whether this ability implies a human-level mastery of CS remains an open question. Understanding the limitations and strengths of LMs can help researchers improve these models, potentially by developing novel ways of integrating external CS knowledge. We devise a series of tests and measurements to systematically quantify their performance on different aspects of CS. We propose the use of cloze testing combined with word embeddings to measure the LM's robustness and confidence. Our results show than although language models tend to achieve human-like accuracy, their confidence is subpar. Future work can leverage this information to build more complex systems, such as an ensemble of symbolic and distributed knowledge.