The INTAMAP project has developed the UncertML candidate standard, which was accepted as an official discussion paper by the OGC in November 2008. UncertML is an XML schema for representing uncertainty in a rigorous manner. Focussing on probabilistic encodings of uncertainty, UncertML provides a representation for any statistic (examples include, mean, median, quantiles, exceedance probabilities, standard deviation, variance, higher order moments), probability distribution (examples include, Gaussian or Normal, Exponential, Gamma, Binomial, Multivariate Normal) or set of realisations. A dictionary is used to describe each UncertML type to maintain the flexibility of the schema. An API is provided (from the UncertML site), which includes support for a range of hard typed examples of commonly used forms.
UncertML has been widely promoted within the OGC, but also to other communities such as the semantic web community and the systems biology community. The development of UncertML will continue beyond the INTAMAP project and will ad support for a range of more complex representations of uncertainty including random functions, Bayes Linear representations and possibly fuzzy approaches. Long term we beleive that UncertML might come to underpin a wide range of web based applications as users increasingly recognise the importance of uncertainty quantification.
UncertML is further described in:
Williams, M., Cornford, D. and Bastin, L., 2008. Describing and Communicating Uncertainty within the Semantic Web, Uncertainty Reasoning for the Semantic Web Workshop, 7th International Semantic Web Conference, 26 October 2008, Karlsruhe, Germany.
Williams, M., Cornford, D., Bastin, L. and Ingram, B. 2008. Exchanging uncertainty: Interoperable geostatistics?, geoENV 2008, 8-10 September 2008, Southampton, UK.