|
Although one must be careful about which labels you look at. In some
cases it appears that labels are part of text (ie sentences) and others look
like pure labels. One would assume that labels would be more pure, as you
say, as they would be unusual words that need to be more clearly
deliniated
****************************** Larry Roux Syracuse University lroux@xxxxxxx ******************************* >>> G.Landini@xxxxxxxxxx 07/11/03 11:07AM >>> Nick & Gordon, On Friday 11 July 2003 15:52, Nick Pelling wrote: > FWIW, I imagine the processing sequence to be more like:- > Galli Romanis oppugnaverunt hastis clamoribusque > --> galch romch opgnavch hasts clamrch& > --> gal chrom cho pgnav chhast sclamr ch& This is interesting. One has to think, though, that this agglomeration should keep Zipf's law more or less unchanged. I am not sure whether agglomeration would do this, because one is creating a new word each time (there should be too many new words, but there aren't). We shouldn't forget about the labels. They are likely to be more accurate (at least as far as the word start and end) than other pieces of text. Cheers, Gabriel ______________________________________________________________________ To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying: unsubscribe vms-list |