Question

I have a termDocumentMatrix created using the tm package in R.

I’m trying to create a matrix/dataframe that has the 50 most frequently occurring terms.

When I try to convert to a matrix I get this error:

> ap.m <- as.matrix(mydata.dtm)
Error: cannot allocate vector of size 2.0 Gb

So I tried converting to sparse matrices using Matrix package:

> A <- as(mydata.dtm, "sparseMatrix") 
Error in as(from, "CsparseMatrix") : 
  no method or default for coercing "TermDocumentMatrix" to "CsparseMatrix"
> B <- Matrix(mydata.dtm, sparse = TRUE)
Error in asMethod(object) : invalid class 'NA' to dup_mMatrix_as_geMatrix

I’ve tried accessing the different parts of the tdm using:

> freqy1 <- data.frame(term1 = findFreqTerms(mydata.dtm, lowfreq=165))
> mydata.dtm[mydata.dtm$ Terms %in% freqy1$term1,]
Error in seq_len(nr) : argument must be coercible to non-negative integer

Here’s some other info:

> str(mydata.dtm)
List of 6
 $ i       : int [1:430206] 377 468 725 3067 3906 4150 4393 5188 5793 6665 ...
 $ j       : int [1:430206] 1 1 1 1 1 1 1 1 1 1 ...
 $ v       : num [1:430206] 1 1 1 1 1 1 1 1 2 3 ...
 $ nrow    : int 15643
 $ ncol    : int 17207
 $ dimnames:List of 2
  ..$ Terms: chr [1:15643] "000" "0mm" "100" "1000" ...
  ..$ Docs : chr [1:17207] "1" "2" "3" "4" ...
 - attr(*, "class")= chr [1:2] "TermDocumentMatrix" "simple_triplet_matrix"
 - attr(*, "Weighting")= chr [1:2] "term frequency" "tf"
> mydata.dtm
A term-document matrix (15643 terms, 17207 documents)

Non-/sparse entries: 430206/268738895
Sparsity           : 100%
Maximal term length: 54 
Weighting          : term frequency (tf)

My ideal output is something like this:

term      frequency
the         2123
and         2095
able         883
...          ...

Any suggestions?

Answer 1

The term-document matrices in tm are already created as sparse matrices. Here, mydata.tdm$i and mydata.tdm$j are the vectors of indexes of the matrix and mydata.tdm$v is the related vector of frequencies. So that you can create a sparse matrix writing :

sparseMatrix(i=mydata.tdm$i, j=mydata.tdm$j, x=mydata.tdm$v)

Then you can use rowSums and link the rows, you’re interested in, to the terms, they stand for, with $Terms.