Problems with Cosine Similarity using library(lsa)

Hi everybody!

I have intended to use library(lsa) on R 64-bits for Windows but it was not possible. Every time I try to launch library(lsa) function R give me back next message:

Loading required package: SnowballError : .onLoad failed in loadNamespace() for 'Snowball', details:  call: NULL  error: .onLoad failed in loadNamespace() for 'rJava', details:  call: stop("No CurrentVersion entry in '", key, "'! Try re-installing Java and make sure R and Java have matching architectures.")  error: objeto 'key' no encontradoError: package ?Snowball? could not be loaded

Of course, I have loaded all necessary packages, but the only way to library(lsa) works it is on R 32-bits release. The problem here is that R don't leave me to load all data from my matrix and tell me that it is not able to load big vectors (may be due to limitations on memory of 32-bit release).

The issue is that I need to calculate cosine similarities on my matrix data. Has somebody any suggestion or idea about how to do it (a different library or a formula to get it)?

Thanks in advance.

Best,

AJ 		 	   		
	[[alternative HTML version deleted]]

Thanks Uwe, you vere right. Now the library is working but I have next problem as well.

I have loaded a matrix (named matrix_v3) from a TXT file. This matrix was previously reached using xtabs function. Now I have used this formula to obtain cosine similarities from my data matrix (following instructions of lsa package instructions):

cosine(matrix_v3, y = NULL)

Finally R returns this message:

Error en cosine(matrix_v3, y = NULL) :   argument mismatch. Either one matrix or two vectors needed as input.

I understand that R don't contemplate matrix_v3 like a real matrix. If this is like this, how can I do? I have tried "as.matrix" but it does not work. Sorry if my questions are not very fine, but I am newbie in using R.

Thanks again.

Date: Thu, 23 Feb 2012 11:35:04 +0100
From: ligges at statistik.tu-dortmund.de
To: anxusgo at hotmail.com
CC: r-help at r-project.org
Subject: Re: [R] Problems with Cosine Similarity using library(lsa)

The error message suggests that you do not have Java installed. And
since you said it works in 32-bit: You only have a 32-bit Java but no
64-bit Java installed in your machine.

Uwe Ligges


On 23.02.2012 11:08, A J wrote:

Hi everybody!

I have intended to use library(lsa) on R 64-bits for Windows but it was not possible. Every time I try to launch library(lsa) function R give me back next message:

Loading required package: SnowballError : .onLoad failed in loadNamespace() for 'Snowball', details:  call: NULL  error: .onLoad failed in loadNamespace() for 'rJava', details:  call: stop("No CurrentVersion entry in '", key, "'! Try re-installing Java and make sure R and Java have matching architectures.")  error: objeto 'key' no encontradoError: package ?Snowball? could not be loaded

Of course, I have loaded all necessary packages, but the only way to library(lsa) works it is on R 32-bits release. The problem here is that R don't leave me to load all data from my matrix and tell me that it is not able to load big vectors (may be due to limitations on memory of 32-bit release).

The issue is that I need to calculate cosine similarities on my matrix data. Has somebody any suggestion or idea about how to do it (a different library or a formula to get it)?

Thanks in advance.

Best,

AJ 		 	   		
	[[alternative HTML version deleted]]

newmatrix<- as.matrix(matrix_v3)

typeof(matrix_v3)