1

ネイティブ R を使用して (tm などの追加のプラグインなしで) ドキュメント ターム マトリックスを作成したいと考えています。データは次のように構成されています。

Doc1: the test was to test the test
Doc2: we did prepare the exam to test the exam
Doc3: was the test the exam
Doc4: the exam we did prepare was to test the test
Doc5: we were successful so we all passed the exam

私が到達したいのは次のとおりです。

         Term Doc1 Doc2 Doc3 Doc4 Doc5 DF
1         all    0    0    0    0    1  1
2         did    0    1    0    1    0  2
3        exam    0    2    1    1    1  4
4      passed    0    0    0    0    1  1
4

1 に答える 1

1

ここにアプローチがありますが、再び tm パッケージを使用しないのはなぜですか?

## Your data
## dat <- structure(list(person = structure(1:5, .Label = c("Doc1", "Doc2", 
##     "Doc3", "Doc4", "Doc5"), class = "factor"), 
##     text = c("the test was to test the test", 
##     "we did prepare the exam to test the exam", "was the test the exam", 
##     "the exam we did prepare was to test the test", 
##     "we were successful so we all passed the exam"
##     )), .Names = c("doc", "text"), class = "data.frame", row.names = c(NA, 
##     -5L))

## Function to turn list of vects into sparse matrix
mtabulate <- function(vects) {
    lev <- sort(unique(unlist(vects)))
    dat <- do.call(rbind, lapply(vects, function(x, lev){ 
        tabulate(factor(x, levels = lev, ordered = TRUE),
        nbins = length(lev))}, lev = lev))
    colnames(dat) <- sort(lev) 
    data.frame(dat, check.names = FALSE)
}


out <- lapply(split(dat$text, dat$doc), function(x) {
    unlist(strsplit(tolower(x), " "))
})

t(mtabulate(out))

##            Doc1 Doc2 Doc3 Doc4 Doc5
## all           0    0    0    0    1
## did           0    1    0    1    0
## exam          0    2    1    1    1
## passed        0    0    0    0    1
## prepare       0    1    0    1    0
## so            0    0    0    0    1
## successful    0    0    0    0    1
## test          3    1    1    2    0
## the           2    2    2    2    1
## to            1    1    0    1    0
## was           1    0    1    1    0
## we            0    1    0    1    2
## were          0    0    0    0    1
于 2013-10-25T15:47:37.820 に答える