在R中的小標題內用空格替換單詞而不進行反連接

Question

我有類似這樣的句子：

小標題：1,782 x 1

Chat
<chr>                                                                                                                                                                    
1 Hi i would like to find out more about the trials
2 Hello I had a guest 
3 Hello my friend overseas right now
...

我想做的是刪除停用詞，例如“ I”，“ hello”。 我已經有了它們的列表，並且想用空格替換這些停用詞。 我嘗試使用mutate和gsub，但它只接受一個正則表達式。 反連接在這里不起作用，因為我正在嘗試使用雙字組/三字組。我沒有一個單詞列來反連接停用詞。

有沒有辦法替換R中每個句子中的所有這些單詞？

Answer 1

我們可以replace標記的嵌套，用空格（ " " ） replace 'stop_words''word'列中找到的'word'，並在按'lines'分組后paste 'word'

library(tidytext)
library(tidyverse)
rowid_to_column(df1, 'lines') %>% 
     unnest_tokens(word, Chat) %>% 
     mutate(word = replace(word, word %in% stop_words$word, " ")) %>% 
     group_by(lines) %>% 
     summarise(Chat = paste(word, collapse=' ')) %>%
     ungroup %>%
     select(-lines)

注意：這會將“ stop_words”數據集中找到的停用詞替換為" "如果我們只需要替換停用詞的自定義子集，則創建這些元素的vector並在mutate步驟中進行更改

v1 <- c("I", "hello", "Hi")
rowid_to_column(df1, 'lines') %>%
  ...
  ...
  mutate(word = replace(word %in% v1, " ")) %>%
  ...
  ...

Answer 2

我們可以使用“ \\\\b停用詞\\\\b ”構造一個模式，然后使用gsub將其替換為“”。 這是一個例子。 請注意，我將ignore.case = TRUE設置為同時包含小寫和大寫字母，但是您可能需要根據需要進行調整。

dat <- read.table(text = "Chat
                  1 'Hi i would like to find out more about the trials'
                  2 'Hello I had a guest' 
                  3 'Hello my friend overseas right now'",
                  header = TRUE, stringsAsFactors = FALSE)

dat
#                                                Chat
# 1 Hi i would like to find out more about the trials
# 2                               Hello I had a guest
# 3                Hello my friend overseas right now

# A list of stop word
stopword <- c("I", "Hello", "Hi")
# Create the pattern
stopword2 <- paste0("\\b", stopword, "\\b")
stopword3 <- paste(stopword2, collapse = "|")

# View the pattern
stopword3
# [1] "\\bI\\b|\\bHello\\b|\\bHi\\b"

dat$Chat <- gsub(pattern = stopword3, replacement = " ", x = dat$Chat, ignore.case = TRUE)
dat
#                                               Chat
# 1     would like to find out more about the trials
# 2                                      had a guest
# 3                     my friend overseas right now