Probing neural language models for understanding of words of estimative probability

TL;DR

This paper investigates whether neural language models understand words of estimative probability (WEP) and their associated probability levels, revealing current models' limitations and potential improvements through fine-tuning.

Contribution

It introduces datasets and tasks to evaluate neural models' grasp of WEP probabilities and probabilistic reasoning, highlighting their current shortcomings and the benefits of fine-tuning.

Findings

Off-the-shelf models fail to accurately predict WEP probabilities.

Fine-tuning improves models' ability to understand and reason with WEP.

Models still struggle with complex probabilistic reasoning tasks.

Abstract

Words of estimative probability (WEP) are expressions of a statement's plausibility (probably, maybe, likely, doubt, likely, unlikely, impossible...). Multiple surveys demonstrate the agreement of human evaluators when assigning numerical probability levels to WEP. For example, highly likely corresponds to a median chance of 0.90+-0.08 in Fagen-Ulmschneider (2015)'s survey. In this work, we measure the ability of neural language processing models to capture the consensual probability level associated to each WEP. Firstly, we use the UNLI dataset (Chen et al., 2020) which associates premises and hypotheses with their perceived joint probability p, to construct prompts, e.g. "[PREMISE]. [WEP], [HYPOTHESIS]." and assess whether language models can predict whether the WEP consensual probability level is close to p. Secondly, we construct a dataset of WEP-based probabilistic reasoning, to test whether language models can reason with WEP compositions. When prompted "[EVENTA] is likely. [EVENTB] is impossible.", a causal language model should not express that [EVENTA&B] is likely. We show that both tasks are unsolved by off-the-shelf English language models, but that fine-tuning leads to transferable improvement.