- From: Kostas Karpouzis <kkarpou@cs.ntua.gr>
- Date: Thu, 23 Nov 2006 14:21:58 +0200
- To: Christian Peter <Christian.Peter@igd-r.fraunhofer.de>
- CC: "Burkhardt, Felix" <Felix.Burkhardt@t-systems.com>, Marc Schröder <schroed@dfki.de>, Jianhua TAO <jhtao@nlpr.ia.ac.cn>, Juan Montero <juancho@die.upm.es>, EMOXG <public-xg-emotion@w3.org>, cpeter@igd-r.fraunhofer.de
Hi Christian, everybody, some (belated) remarks: - 'purpose of classification': if this is related to scope, e.g. high- or low-level, automatic/manual, etc. then I would think that it's not very important, since it can be deduced by the nature and resolution of the detected features - 'ID of labeller': possible ethics/data protection issue here - 'time span'/'atomic entity': possibly *one* of them is needed - 'could it be broken down' assumes recognition/identification of emotion, which is different to annotation In general, the 'links to pointers' and 'meta info' sections are nice to have, but tend to move the annotation result away from a standardized form. Hence, I would prefer any reference to 'external event' or context to be in the 'main body' of the annotation; could live without it, though :)) Best regards, Kostas P.S. Thanks for the great work, Christian! Christian Peter wrote: > Hi folks, namely Felix, Marc, Jianhua, Juan, and Kostas! > > Have you found time to have a look at the UC2 requirements I compiled > from your use cases? > > Please do me the favour and tick one of the following: > > [] This list is perfect > [] let's talk about it on Monday, 27 November > [] I have comments, will send them soon > [] give me some more days (phone meeting 11. Dec.) > > Thank you so much, > > Christian
Received on Thursday, 23 November 2006 12:27:14 UTC