将 DocumentTermMatrix 转换为 dgTMatrix

Question

我正在尝试通过 text2vec 的 LDA 实现从 tm-package 运行 AssociatedPress 数据集。

我面临的问题是数据类型的不兼容：AssociatedPress 是 tm::DocumentTermMatrix，而 tm::DocumentTermMatrix 又是 slam::simple_triplet_matrix 的子类。 text2vec 但是希望 x 到 text2vec::lda$fit_transform(x = ...) 的输入为 Matrix::dgTMatrix。

因此我的问题是：有没有办法将DocumentTermMatrix 强制为text2vec 接受的东西？

最小（失败）示例：

library('tm')
library('text2vec')

data("AssociatedPress", package="topicmodels")

dtm <- AssociatedPress[1:10, ]

lda_model = LDA$new(
  n_topics = 10,
  doc_topic_prior = 0.1,
  topic_word_prior = 0.01
)

doc_topic_distr =
  lda_model$fit_transform(
    x = dtm,
    n_iter = 1000,
    convergence_tol = 0.001,
    n_check_convergence = 25,
    progressbar = FALSE
  )

...给出：

base::rowSums(x, na.rm = na.rm, dims = dims, ...) : 'x' 必须是 至少二维数组

Answer 1

答案在@Dmitriy Selivanov 提供的duplicate 中。但是它没有提到它来自基础包Matrix。

由于我没有安装topicmodels，我将使用crude 数据集，它包含在tm 包中。原理是一样的。

library(tm)
data("crude")

dtm <- DocumentTermMatrix(crude,
                          control = list(weighting =
                                           function(x)
                                             weightTfIdf(x, normalize =
                                                           FALSE),
                                         stopwords = TRUE))

# transform into a sparseMatrix dgcMatrix
m <-  Matrix::sparseMatrix(i=dtm$i, 
                           j=dtm$j, 
                           x=dtm$v, 
                           dims=c(dtm$nrow, dtm$ncol),
                           dimnames = dtm$dimnames)
str(m)
Formal class 'dgCMatrix' [package "Matrix"] with 6 slots
  ..@ i       : int [1:1890] 6 1 18 6 6 5 9 12 9 5 ...
  ..@ p       : int [1:1201] 0 1 2 3 4 5 6 8 9 11 ...
  ..@ Dim     : int [1:2] 20 1200
  ..@ Dimnames:List of 2
  .. ..$ Docs : chr [1:20] "127" "144" "191" "194" ...
  .. ..$ Terms: chr [1:1200] "\"(it)" "\"demand" "\"expansion" "\"for" ...
  ..@ x       : num [1:1890] 4.32 4.32 4.32 4.32 4.32 ...
  ..@ factors : list()

其余代码：

library(text2vec)

lda_model <- LDA$new(
  n_topics = 10,
  doc_topic_prior = 0.1,
  topic_word_prior = 0.01
)

doc_topic_distr <-
  lda_model$fit_transform(
    x = m,
    n_iter = 1000,
    convergence_tol = 0.001,
    n_check_convergence = 25,
    progressbar = FALSE
  )

INFO [2018-04-15 10:40:00] iter 25 loglikelihood = -32949.882
INFO [2018-04-15 10:40:00] iter 50 loglikelihood = -32901.801
INFO [2018-04-15 10:40:00] iter 75 loglikelihood = -32922.208
INFO [2018-04-15 10:40:00] early stopping at 75 iteration