使用tm_map的自定义函数来维护用户定义的元数据

Question

我有一个函数，可用于根据键/值字典转换令牌。

dictionary <- c("casa", "barco", "carro", "arbol")
names(dictionary) <- c("home", "boat", "car", "tree")

translate2 <- function (text, dictionary) {
  text_out <- character(0)
  for (i in 1:length(text)) {
    text.split <- strsplit(text[i], "\\s")
    translation <- dictionary[unlist(text.split)]
    text_out <- append(text_out, paste(translation, sep="", collapse=" "))
  }
  PlainTextDocument(text_out, id = ID(text), author = Author(text))
}

此功能对于meta`Author可以正常使用：

library(tm)

text <- "My car is on the tree next to my home under the boat"
corpus <- Corpus(VectorSource(text))
meta(corpus, "Author", type="local") <- "Kant"
meta(corpus, "TextID", type="local") <- "121212"
meta(corpus[[1]], "Author")
# [1] "Kant"

corpus <- tm_map(corpus, translate2, dictionary)
meta(corpus[[1]], "Author")
# [1] "Kant" 
corpus[[1]]
# NA carro NA NA NA arbol NA NA NA casa NA NA barco

但是，当我尝试通过函数的稍微修改的版本传递用户定义的元数据（例如TextID ，

translate1 <- function (text, dictionary) {
  text_out <- character(0)
  for (i in 1:length(text)) {
    text.split <- strsplit(text[i], "\\s")
    translation <- dictionary[unlist(text.split)]
    text_out <- append(text_out, paste(translation, sep="", collapse=" "))
  }
  PlainTextDocument(text_out, id = ID(text), author = Author(text), 
                    TextID = TextID(text))
} 

我懂了

text <- "My car is on the tree next to my home under the boat"
corpus <- Corpus(VectorSource(text))
meta(corpus, "Author", type="local") <- "Kant"
meta(corpus, "TextID", type="local") <- "121212"
meta(corpus[[1]], "Author")
# [1] "Kant"
meta(corpus[[1]], "TextID")
# [1] "121212"

corpus <- tm_map(corpus, translate1, dictionary)
# Error in PlainTextDocument(text_out, id = ID(text), author = Author(text),  : 
#                              unused argument (TextID = TextID(text)) 

Answer 1

您的方法存在一些问题：

1. PlainTextDocument没有参数TextID （这导致了您的错误）
2. 没有名为TextID函数

从?PlainTextDocument看来，您正在寻找的参数称为localmetadata 。

这是translate1的一个版本，该版本似乎可以正常工作：

translate1 <- function (text, dictionary) {
  text_out <- character(0)
  for (i in 1:length(text)) {
    text.split <- strsplit(text[i], "\\s")
    translation <- dictionary[unlist(text.split)]
    text_out <- append(text_out, paste(translation, sep="", collapse=" "))
  }
  PlainTextDocument(text_out, id = ID(text), author = Author(text), 
                    localmetadata = list(TextID = meta(text, "TextID")))
} 

text <- "My car is on the tree next to my home under the boat"
corpus <- Corpus(VectorSource(text))
meta(corpus, "Author", type="local") <- "Kant"
meta(corpus, "TextID", type="local") <- "121212"
meta(corpus[[1]], "Author")
# [1] "Kant"
meta(corpus[[1]], "TextID")
# [1] "121212"

corpus <- tm_map(corpus, translate1, dictionary)
meta(corpus[[1]], "Author")
# [1] "Kant"
meta(corpus[[1]], "TextID")
# [1] "121212"
corpus[[1]]
# NA carro NA NA NA arbol NA NA NA casa NA NA barco