In Latent Semantic Analysis, we get 3 matrices from the singular value decomposition (SVD), but I am confused - which matrix do we use for cosine similarity?
Asked
Active
Viewed 1,913 times
1
-
1I do not vote this to close because for me it is very clear what the OP is asking. Having said that, I would suggest to the OP to actually write out what LSA is doing. I am pretty certain he will be able to answer his own question immediately afterwards; in particular he wants to examine the relation between PCA and SVD. – usεr11852 May 07 '15 at 07:47
-
This question may also be useful: http://stats.stackexchange.com/questions/134282/relationship-between-svd-and-pca-how-to-use-svd-to-perform-pca – Alexey Grigorev May 07 '15 at 09:15