Hedges are linguistic constructions used to modify or contextualise the meanings and implications of utterances. Numerical hedges, called approximators, are special kinds of hedges used with numbers, such as 'about', 'around', 'nearly', or 'at least'.
Numerical hedging is important in risk analysis because we need to understand what people mean when they use hedges in numerical expressions. If "about 5.43" is not the same as "5.43", just how is it different? How should we quantitatively interpret numerical expressions that include such hedges?
We would also like to know whether and how we should use hedges when we try to express uncertainty about numerical results. How should we design automated systems intended to inform people about quantitative results so that the numerical expressions are correctly understood? In particular, how can we use hedges to communicate uncertainty effectively?
Previous work in quantifying numerical hedges is described at https://sites.google.com/site/numericalhedging/.
Dominik Hose reanalysed the data to these obtain consonant possibility structures for various hedges:
BLUE: Melting Transform
ORANGE: Imprecise Probabilities-to-Possibility Transform based on Melting Transform