Natural Language Understanding (NLU) - non-Alphanumeric characters when adding synonymsIssue Some users might find difficulties when trying to add non-dictionary words as a synonym to the NLU, such as words with a hyphen "-". Currently, only alphanumeric characters are accepted and if anyone tries to do otherwise, the following error message will be displayed: ReleaseNew York +CauseVocabulary is a way of converting an unknown word into a known word, so the synonym should always be a valid word known to the NLU Model Builder. However, in some cases the NLU Model Builder understands non-alphanumeric words. ResolutionThe vocabulary is used by the NLU Model Builder when you submit your NLU model for training. In this scenario, the baseword "M-Desktop" is an unknown vocabulary word and a recognised vocabulary word needs to be added, so that the NLU Model Builder can work with the baseword. For example, replacing "M-Desktop" by "mdesktop" or "MDesktop" would still be an unrecognised vocabulary word. Therefore, the solution is to replace “M-Desktop” or “m-desktop” with “virtual desktop” (a recognised word) as a synonym to the baseword, which the NLU Model Builder can work with and we then know what the baseword means during the inference. Once a recognised word has been added as a synonym, the user can type “M-Desktop” in the Virtual Agent chat window and since NLU now knows what this word means, we convert it back to “virtual desktop” and do the inference, allowing Virtual Agent users to enter non-alphanumeric characters in the Virtual Agent chat window that would be recognised. Related LinksResolve an unknown word [Orlando]