I'm working with the cluster analysis algorithm (EM) in SQL 2005. I have tried to find documentation on the function PredictCaseLikelihood without luck. Is there any reference on how this function is defined?
View 5 RepliesI am working on a text mining application wherein I need to detect unusual/anomalous sentences in text. Certain sentences, that I know occur very frequently, are given a likelihood of 0.2 by PredictCaseLikelihood. Other sentences that are just as frequent get a much higher likelihood (>0.9). I am using the NORMALIZED option. The only significant difference between these sentences is their length. The one with the lower likelihood has only 2 words in it, whereas the one with the higher likelihood has more than 10 words. The problem is that the shorter sentences end up being interpreted as anomalous, when in fact they are'nt. Any suggestions?
View 2 Replies View Related