Dear Jim, That should certainly have been a qualified "no". Original data are the best for validation. For example, a Chinese address is basically Chinese characters and that is the primary form. If data have been transformed, the transformation has to be high quality (i.e. probably not automated except possibly for some alphabetic scripts) and a check needs to be made that the transformed data match the original data. Regards, Chris. From: James Galvin [mailto:jgalvin@afilias.info] Sent: 17 July 2014 14:17 To: Dillon, Chris; gnso-contactinfo-pdp-wg@icann.org Subject: Re: [gnso-contactinfo-pdp-wg] Transformation required at the validation stage? On 7/14/14, 11:46 AM, Dillon, Chris wrote: At one point Jim Galvin asked whether transformation would be required at the validation stage. I would say no to that question. The original language/script data are primary and suitable for validation. It's not that transformed data cannot be validated, but they need to be high quality (outside Greek, Cyrillic and alphabetic scripts often created by human being rather than a computer). Actually I think Rudi made a similar point later in the call, but just to clarify that. Perhaps I should listen to the recording to get more context. However, one point that occurs to me is that I don't think you can give an unequivocal no just yet. Consider the question of whether or not a single script and language is required. If we decide that all registration information is to be in a single script and language, then we can consider the question of which presentation of the data to validate: the original input data or the transformed data. I would argue that if the chosen single script and language is to be "official", then that presentation of the data should be subject to all the validation requirements. The input data will need to be normalized so that it can be transformed. We will need to keep the input form of the data for audit purposes. However, validating the input form seems to me to be tantamount to simply supporting all languages and scripts, which makes me wonder why we would choose a single language and script for all data. This is my current thinking. Jim Regards, Chris. -- Research Associate in Linguistic Computing, Centre for Digital Humanities, UCL, Gower St, London WC1E 6BT Tel +44 20 7679 1599 (int 31599) www.ucl.ac.uk/dis/people/chrisdillon<http://www.ucl.ac.uk/dis/people/chrisdillon>