X-Google-Language: ENGLISH,ASCII-7-bit X-Google-Thread: f996b,a634ddb4f5113d2a X-Google-Attributes: gidf996b,public From: Ilmari Karonen Subject: Re: Warning -- ASCII art and Netscape Messenger 5.0 Date: 1999/11/09 Message-ID: <809i4o$j48$1@tron.sci.fi>#1/1 X-Deja-AN: 546496240 Distribution: world References: <3825403D.D240CF94@student.canterbury.ac.nz> <382671CB.8D212EA4@bucksch.org> <8095ka$qq0$2@news.inet.tele.dk> Organization: (dis)Order of the Holy Spoon (or whatever) User-Agent: postit.pl 0.03 Newsgroups: alt.ascii-art On Tue, 9 Nov 1999 13:25:52 +0100, Christian 'CeeJay' Jensen (REMOVEnewsgroup@CeeJay.findhere.com) wrote: : btw how would you catch all the hundreds of smilies out there .. : :) 8) =8) >8) >;> 8:-) :D ]|> @:) 0:) %} just to name a few .. : Would you do a grapic for ALL of them ? http://www.deja.com/getdoc.xp?AN=312573064&fmt=text Message-ID: <68j7dn$ccq@tron.sci.fi> While that post mostly deals with composing emoticons instead of parsing them, I think the following quote is relevant to the current discussion: > As an exercise for readers, try to devise an algorithm to identify > emoticons as well as possible. Consider for example the following > points: > > * Emoticons may use other characters than "()" as a mouth. "[]{}<>|/\&" > and so on should be easy, but what about :-X , :-9 or :-I ? > * What about letters in other parts of an emoticon? How would you parse > "..about the exam (I got a B-) which we.."? How about "Bob" (a smiley > with sunglasses, potato nose and tongue sticking out ;) )? > * What about emoticons with (parts of) a body, such as the (in)famous > "big girl smiley"? I think it should be obvious that there can be no foolproof way to detect emoticons - what is or is not an emoticon depends ultimately on context, making the problem equally difficult to natural language parsing. The Jargon File calls such deceptively simple problems is "AI-complete", as solving the general case would require true conscious thinking. -- Ilmari Karonen http://www.sci.fi/~iltzu/