Hi,
I don't know something about this special implementation, but maybe you
are interested in the following urls:
Wikipedia:
http://en.wikipedia.org/wiki/Likelihood
Paper from Ted Dunning
'Accurate Methods for the Statistics of Surprise and Coincidence'
http://www.aclweb.org/anthology/J/J93/J93-1003.pdf
Cheers, Walter
Cui tony schrieb:
Hi,
I read this function in the source code of taste, and I have some
questions on the algorithm of similarity calculation:
public double itemSimilarity(long itemID1, long itemID2) throws
TasteException {
int preferring1and2 = dataModel.getNumUsersWithPreferenceFor(itemID1,
itemID2);
if (preferring1and2 == 0) {
return Double.NaN;
}
int preferring1 = dataModel.getNumUsersWithPreferenceFor(itemID1);
int preferring2 = dataModel.getNumUsersWithPreferenceFor(itemID2);
int numUsers = dataModel.getNumUsers();
double logLikelihood =
twoLogLambda(preferring1and2, preferring1 - preferring1and2,
preferring2, numUsers - preferring2);
return 1.0 - 1.0 / (1.0 + logLikelihood);
}
static double twoLogLambda(double k1, double k2, double n1, double n2) {
double p = (k1 + k2) / (n1 + n2);
return 2.0 * (logL(k1 / n1, k1, n1) + logL(k2 / n2, k2, n2) - logL(p,
k1, n1) - logL(p, k2, n2));
}
Is there any academic paper on this function? Why we should calculate the
similarity by the upon formula?
Thanks!
--
WeFind - Genau was Du suchst
Die erste Web 2.0 Suchmaschine jetzt auf www.wefind.de und im iTunes
AppStore für iPhone und iPod Touch!
--------------------------------
Walter Tietze
Senior Softwareentwickler
Softwareentwicklung
neofonie
Technologieentwicklung und
Informationsmanagement GmbH
Robert-Koch-Platz 4
10115 Berlin
fon: +49.30 24627 318
fax: +49.30 24627 120
[email protected]
http://www.neofonie.de
Handelsregister
Berlin-Charlottenburg: HRB 67460
Geschaeftsfuehrung
Helmut Hoffer von Ankershoffen
(Sprecher der Geschaeftsfuehrung)
Nurhan Yildirim
--------------------------------
